Tag Archives: video element

Summary Video Accessibility Talk

I’ve just got off a call to the UK Digital TV Group, for which I gave a talk on HTML5 video accessibility (slides best viewed in Google Chrome).

The slide provide a high-level summary of the accessibility features that we’ve developed in the W3C for HTML5, including:

  • Subtitles & Captions with WebVTT and the track element
  • Video Descriptions with WebVTT, the track element and speech synthesis
  • Chapters with WebVTT for semantic navigation
  • Audio Descriptions through synchronising an audio track with a video
  • Sign Language video synchronized with a main video

I received some excellent questions.

The obvious one was about why WebVTT and not TTML. While for anyone who has tried to implement TTML support, the advantages of WebVTT should be clear, for some the decision of the browsers to go with WebVTT still seems to be bothersome. The advantages of CSS over XSL-FO in a browser-context are obvious, but not as much outside browsers. So, the simplicity of WebVTT and the clear integration with HTML have to speak for themselves. Conversion between TTML and WebVTT was a feature that was being asked for.

I received a question about how to support ducking (reduce the volume of the main audio track) when using video descriptions. My reply was to either use video descriptions with WebVTT and do ducking during the times that a cue is active, or when using audio descriptions (i.e. actual audio tracks) to add an additional WebVTT file of kind=metadata to mark the intervals in which to do ducking. In both cases some JavaScript will be necessary.

I received another question about how to do clean audio, which I had almost forgotten was a requirement from our earlier media accessibility document. “Clean audio” consists of isolating the audio channel containing the spoken dialog and important non-speech information that can then be amplified or otherwise modified, while other channels containing music or ambient sounds are attenuated. I suggested using the mediagroup attribute to provide a main video element (without an audio track) and then the other channels as parallel audio tracks that can be turned on and off and attenuated individually. There is some JavaScript coding involved on top of the APIs that we have defined in HTML, but it can be implemented in browsers that support the mediagroup attribute.

Another question was about the possibilities to extend the list of @kind attribute values. I explained that right now we have a proposal for a new text track kind=”forced” so as to provide forced subtitles for sections of video with foreign language. These would be on when no other subtitle or caption tracks are activated. I also explained that if there is a need for application-specific text tracks, the kind=”metadata” would be the correct choice.

I received some further questions, in particular about how to apply styling to captions (e.g. color changes to text) and about how closely the browser are able to keep synchronization across multiple media elements. The earlier was easily answered with the ::cue pseudo-element, but the latter is a quality of implementation feature, so I had to defer to individual browsers.

Overall it was a good exercise to summarize the current state of HTML5 video accessibility and I was excited to show off support in Chrome for all the features that we designed into the standard.

3rd W3C Web and TV Workshop, Hollywood

Curious about any new requirements that the TV community may have for HTML5 video, I attended the W3C Web and TV Workshop in Hollywood last week. It’s already the third of its kind and was also the largest to date showing an increasing interest of the TV community to converge with the Web community.

The Workshop Aim

I went into the Workshop not quite knowing what to expect. My previous contact with members of this community was restricted to email exchanges on the W3C Web and TV Interest Group (IG) mailing list. I knew there was some interest in video accessibility (well: particularly captions) and little knowledge of existing HTML5 specifications around text tracks and why the browsers were going with WebVTT. So I had decided to attend the workshop to get a better understanding of the community, it’s background, needs, and issues, and to hopefully teach some of the ways of HTML5. For that reason I had also submitted a WebVTT presentation/demo.

As it turned out, the workshop had as its key target the facilitation of communication between the TV and the HTML5 community. The aim was to identify features that need to be added to the HTML5 video element to satisfy the needs of the TV community. I obviously came to the right workshop.

The process that is being used by the W3C in the Interest Group is to have TV community members express their needs, then have HTML5 experts express how these needs can be satisfied with existing HTML5 features, then make trial implementations and identify any shortcomings, then move forward to progress these through HTML5 or HTML.next. This workshop clearly focused on the first step: expressing needs.

Often times it was painful for me to watch presenters defending their requirements and trying to impress on the audience how important a certain feature is to them when that features actually already has a HTML5 specification, but just not yet a browser implementations. That there were so few HTML5 video experts present and that they were given very little space to directly reply to the expressed needs and actually explain what is already possible (or specified to be possible) was probably one of the biggest drawbacks of the workshop.

To be fair, detailed technical discussions were not possible in a room with 150 attendees with a panel sitting at the front discussing topics and taking questions. Solving a use case with existing HTML5 markup and identifying the gaps requires smaller break-out groups of a maximum of maybe 20 people and sufficient HTML5 knowledge in the room. Ultimately they require a single person to try to implement it using JavaScript alone, and, failing that, writing browser extensions. Only such code actually proves that a feature is missing.

Now, the video features of HTML5 are still continuing to change almost on a daily basis. Much development is, for example, happening around real-time communication features and around the track element as we speak. So, focusing on further requirements finding around HTML5 video for now is probably a good thing.

The TV Community Approach

Before I move on to some of the topics covered by the workshop, I have to express some concern about the behaviour that I observed with lots of the TV community folks. Many people tried pushing existing solutions from other spaces into the Web unchanged with a claim of not re-inventing the wheel and following paved cowpaths, which are some of the underlying design principles for HTML5. I can understand where such behaviour originates thinking that having solved the same problems elsewhere before, those solutions should apply here, too. But I would like to warn people of this approach.

If we blindly apply solutions that were not developed for HTML5 into HTML we will end up with suboptimal solutions that will hurt us further down the track. The principles of not re-inventing the wheel and following paved cowpaths were introduced for features that were already implemented by browsers or in de-facto standard use by JavaScript libraries. They were not created for new features in HTML. The video element is a completely new feature in HTML thus everything around it is new.

I would therefore like to see some more respect given to HTML5 and the complexities involved in finding the best possible technical solutions for the Web given that the video element does not stand alone in HTML5, but is part of a much larger picture of technical capabilities on the Web where many of the requested features for TV applications may already be solved by existing HTML markup that is not part of the video element.

Also, HTML5 is not just about the HTML markup, but also about CSS and JavaScript and HTTP. There are several layers of technology involved in creating a Web application: not only a separation of work between client and servers, but also between the Operating System, the media framework, the browser, browser plugins, and JavaScript has to be balanced. To get this balance right is a fine art that will take many discussion, many experiments and sometimes several design approaches. We need patience and calm to work through this, not a rushed adoption of existing solutions from other spaces.

New Requirements

Now let’s get to the take-aways I had from the workshop’s sessions:

Session 1 / Content Provider and Consumer Perspective:

The sessions participants postulate that we will see the creation of application stores for TV applications similar to how we have experienced this for mobile phones and tablets. People enjoy collecting apps like they collect badges. Right now, the app store domain is dominated by native apps and now Web apps. The reason is that we haven’t got a standard platform for setting up Web app stores with Web apps that work in all browsers on all operating systems. Thus, developers have to re-deploy their app for many environments.

While essentially an orthogonal need to HTML standardisation, this seems to be one of the key issues that keep Web apps back from making big market inroads and W3C may do well in setting up a new WG to define a standard Web app manifest format and JS APIs.

Session 2+3 / Multi-screen TV in the Home Network:

Several technologies of hybrid TV broadcast and set-top-box Web content delivery were being pointed out, including the European HbbTV and the Japanese Hybridcast, the latter of which gave an in-depth demo.

Web purists would probably say that it would be simpler to just deliver all content over the Web and not have to worry about any further technical challenges encountered by having to synchronize content received via two vastly different delivery mechanisms. I personally believe this development is one of business models: we don’t yet know exactly how to earn money from TV content delivered over the Internet, but we do know how to do so with TV content. So, hybrids allow the continuation of existing income streams while allowing the features to be augmented with those people enjoy from the Internet.

Should requirements that emerge from such a use case for HTML5 video be taken seriously? I think they absolutely should. What I see happening is that a new way of using the Web is starting to emerge. The new way is video-focused rather than text-focused. We receive our Web content by watching video programming online – video channels, not Web pages are the core content that we consume in the living room. Video channels are where we start our browsing experience from. Search may still be our first point of call, but it will be search for video content or a video-centric app rather than search for a Web site.

And it will be a matter of many interconnected devices in the house that contribute to the experience: the 5.1 stereos that are spread all over the house and should receive our video’s sound, the different screens in the different areas of our house between which we move around, and remote controls, laptops or tablets that function as remote controls and preview stations and are used to determine our viewing experience and provide a back-channel to the publishers.

We have barely begun to identify how such interconnected devices within a home fit within the server-client-based view of the Web world, and the new Web Sockets functionality. The Home Networking Task Force of the Web and TV IG is looking at the issues and analysing existing protocols and standards that solve this picture. But I have a gnawing feeling that the best solution will be something new that is more Web-specific and fits better with the technology layers of the Web.

Session 4 / Synchronized Metadata:

The TV environment offers many data services, some of which have been legally prescribed. This session analysed TV needs and how they can be satisfied with current HTML5.

Subtitles and closed captioning support are one of the key requirements that have been legally prescribed to allow for equal access of non-native speakers, and blind and vision-impaired users to TV content. After demonstration of some key features defined into the HTML5 track element and the WebVTT format, it was generally accepted that HTML5 is making big progress in this space, in particular that browsers are in the process of implementing support for the track element. A concern still exists for complete coverage of all the CEA-608/708 features in WebVTT.

Further concern was raised for support of audio descriptions and audio translations, in particular since no browser has as yet committed to implementing the HTML5’s media multitrack API with the @mediagroup attribute. In this context I am excited to see first JavaScript polyfills emerge (see captionator.js & mediagroup.js).

Another concern was that many captions are actually delivered as raster images (in particular DVD captions) and how that would work in the Web context. The proposal was to use WebVTT and encode the raster images as data-URIs included in timed cues, then render them by JavaScript as an overlay. This is something to explore further.

Demos were shown using WebVTT to synchronize ads with videos, to display related metadata from a user’s life log with videos, to display thumbnails along a video’s timeline, and to show the rendering of text descriptions through screen readers. General agreement by the panel was that WebVTT offers many opportunities and that this area will continue to need further development and that we will see new capabilities on the Web around metadata that were not previously possible on TV.

Session 5 / Content Format and Codecs: DASH and Codec standards

The introduction of HTTP adaptive streaming into HTML5 was one of the core issues that kept returning in the discussions. This panel focused on MPEG DASH, but also mentioned the need for programmatic implementation of adaptive streaming functionality.

The work around MPEG DASH would require specifications of how to use DASH with WebM and Ogg Theora, as well as a specification of a HTML5 profile for DASH, which would limit the functionality possible in DASH files to the ones needed in a HTML5 video element. One criticism of DASH was its verbosity. Another was its unclear patent position. Panel attendees with included Qualcomm, Apple and Microsoft made very clear that their position is pro a royalty-free use of DASH.

The work around a programmatic implementation for adaptive streaming would require at least a JavaScript API to measure the quality of service of a presented video element and a JavaScript API to feed the video element with chunks of (encrypted) video content on the fly. Interestingly enough, there are existing experiments both around Video metrics and MediaSource extensions, so we can expect some progress in this space, even if these are not yet a strong focus of the HTML WG.

I would personally support the creation of Community Group at the W3C around HTTP adaptive streaming and DASH. I think it would work towards alleviating the perceived patent issues around DASH and allow the right members of the community to participate in preparing a specification for HTML5 without requiring them to become W3C members.

Session 6 / Content Protection and DRM

A core concern of the TV community is around content protection. The requirements in this space seem, however, very confused.

The key assumption here is that Web browsers should support the decoding of DRM-protected content in the HTML5 video element because the video element provides a desirable JavaScript API, accessibility features (the track element), default controls, and the possibility to synchronize multiple media elements. However, at the same time, the video element is part of the core content of a Web page and thus allows direct access to the image content in a canvas etc, so some of its functionality is not desirable.

The picture is further confused by requests for authentication, authorization, encryption, obfuscation, same-origin, secure transmission, secure decryption key delivery, unique content identification and other “content protection” techniques without a clear understanding of what is already possible on the Web and what requirements to content publishers actually have for delivering their content on the Web. This is further complicated by the fact that there are many competing solutions for DRM systems in the market with no clear standard that all browsers could support.

A thorough analysis of the technologies and solutions available in this space as well as an analysis of the needs for HTML5 is required before it becomes clear what solution HTML5 browsers may need to support. There seemed to be agreement in the group, though, that browsers would not need to implement DRM solutions, but rather only hand through the functionality of the platform on which they are running (including the media frameworks and operating system functionalities). How this is supposed to work was, however, unclear.

Session 7 / Web & TV: Additional Device & User Requirements

This was a catch-all session for topics that had not been addressed in other sessions. Among the topics addressed in this group were:

  • Parental Guidance: how to deal with ratings in an internationally inconsistent ratings landscape, how to deliver the ratings with the content, and how to enforce the viewing restrictions
  • Emergency Notifications: how to replicate on the Web the emergency notification functionality of TV by providing text overlays to alert users
  • TV channels: how to detect what channels of programming are available to users

Overall, the workshop was a worthwhile experience. It seems there is a lot of work still ahead for making HTML5 video the best it can be on the Web.

Recent developments around WebVTT

People have been asking me lots of questions about WebVTT (Web Video Text Tracks) recently. Questions about its technical nature such as: are the features included in WebVTT sufficient for broadcast captions including positioning and colors? Questions about its standardisation level: when is the spec officially finished and when will it move from the WHATWG to the W3C? Questions about implementation: are any browsers supporting it yet and how can I make use of it now?

I’m going to answer all of these questions in this post to make it more efficient than answering tweets, emails, and skype and other phone conference requests. It’s about time I do a proper post about it.

Implementations

I’m starting with the last area, because it is the simplest to answer.

No, no browser has as yet shipped support for the <track> element and therefore there is no support for WebVTT in browsers yet. However, implementations are in progress. For example, Webkit has recently received first patches for the track element, but there is still an open bug for a WebVTT parser. Similarly, Firefox can now parse the track element, but is still working on the element’s actual functionality.

However, you do not have to despair, because there are now a couple of JavaScript polyfill libraries for either just the track element or for video players with track support. You can start using these while you are waiting for the browsers to implement native support for the element and the file format.

Here are some of the libraries that I’ve come across that will support SRT and/or WebVTT (do leave a comment if you come across more):

  • Captionator – a polyfill for track and SRT parsing (WebVTT in the works)
  • js_videosub – a polyfill for track and SRT parsing
  • jscaptions – a polyfill for track and SRT parsing
  • LeanBack player – a video player with track and SRT, SUB, DFXP, and soon full WebVTT parsing support
  • playr – a video player that includes track and WebVTT parsing
  • MediaElementJS – a video player that includes track and SRT parsing
  • Kaltura’s video player – a video player that includes track and SRT parsing

I am actually most excited about the work of Ronny Mennerich from LeanbackPlayer on WebVTT, since he has been the first to really attack full support of cue settings and to discuss with Ian, me and the WHATWG about their meaning. His review notes with visual description of how settings are to be interpreted and his demo will be most useful to authors and other developers.

Standardisation

Before we dig into the technical progress that has been made recently, I want to answer the question of “maturity”.

The WebVTT specification is currently developed at the WHATWG. It is part of the HTML specification there. When development on it started (under its then name WebSRT), it was also part of the HTML5 specification of the W3C. However, there was a concern that HTML5 should be independent of the chosen captioning format and thus WebVTT currently only exists at the WHATWG.

In recent months – and particularly since browser vendors have indicated that they will indeed implement support for WebVTT as their implementation of the <track> element – the question of formal standardization of WebVTT at the W3C has arisen. I’m involved in this as a Google contractor and we’ve put together a proposed charter for a WebVTT Working Group at the W3C.

In the meantime, standardization progresses at the WHATWG productively. Much feedback has recently been brought together by Ian and changes have been applied or at least prepared for a second feature set to be added to WebVTT once the first lot is implemented. I’ve captured the potentially accepted and rejected new features in a wiki page.

Many of the new features are about making the WebVTT format more useful for authoring and data management. The introduction of comments, inline CSS settings and default cue settings will help authors reduce the amount of styling they have to provide. File-wide metadata will help with the exchange of management information in professional captioning scenarios and archives.

But even without these new features, WebVTT already has all the features necessary to support professional captioning requirements. I’ve prepared a draft mapping of CEA-608 captions to WebVTT to demonstrate these capabilities (CEA-608 is the TV captioning standard in the US).

So, overall, WebVTT is in a great state for you to start implementing support for it in caption creation applications and in video players. There’s no need to wait any longer – I don’t expect fundamental changes to be made, but only new features to be added.

New WebVTT Features

This takes us straight to looking at the recently introduced new features.

  • Simpler File Magic:
    Whereas previously the magic file identifier for a WebVTT file was a single line with “WEBVTT FILE”. This has now been changed to a single line with just “WEBVTT”.
  • Cue Bold Span:
    The <b> element has been introduced into WebVTT, thus aligning it somewhat more with SRT and with HTML.
  • CSS Selectors:
    The spec already allowed to use the names of tags, the classes of <c> tags, and the voice annotations of <v> tags as CSS selectors for ::cue. ID selector matching is now also available, where the cue identifier is used.
  • text-decoration support:
    The spec now also supports the CSS text-decoration property for WebVTT cues, allowing functionality such as blinking text and bold.

Further to this, the email identifies the means in which WebVTT is extensible:

  • Header area:
    The WebVTT header area is defined through the “WEBVTT” magic file identifier as a start and two empty lines as an end. It is possible to add into this area file-wide information header information.
  • Cues:
    Cues are defined to start with an optional identifier, and then a start/end time specification with “–>” separator. They end with two empty lines. Cues that contain a “–>” separator but don’t parse as valid start/end time are currently skipped. Such “cues” can be used to contain inline command blocks.
  • Inline in cues:
    Finally, within cues, everything that is within a “tag”, i.e. between “”, and does not parse as one of the defined start or end tags is ignored, so we can use these to hide text. Further, text between such start and end tags is visible even if the tags are ignored, so wen can introduce new markup tags in this way.

Given this background, the following V2 extensions have been discussed:

  • Metadata:
    Enter name-value pairs of metadata into the header area, e.g.

    WEBVTT
    Language=zh
    Kind=Caption
    Version=V1_ABC
    License=CC-BY-SA
    
    1
    00:00:15.000 --> 00:00:17.950
    first cue
  • Inline Cue Settings:
    Default cue settings can come in a “cue” of their own, e.g.

    WEBVTT
    
    DEFAULTS --> D:vertical A:end
    
    00:00.000 --> 00:02.000
    This is vertical and end-aligned.
    
    00:02.500 --> 00:05.000
    As is this.
    
    DEFAULTS --> A:start
    
    00:05.500 --> 00:07.000
    This is horizontal and start-aligned.
    
  • Inline CSS:
    Since CSS is used to format cue text, a means to do this directly in WebVTT without a need for a Web page and external style sheet is helpful and could be done in its own cue, e.g.

    WEBVTT
    
      STYLE -->
      ::cue(v[voice=Bob]) { color: green; }
      ::cue(c.narration) { font-style: italic; }
      ::cue(c.narration i) { font-style: normal; }
    
      00:00.000 --> 00:02.000
      <v Bob>Welcome.
    
      00:02.500 --> 00:05.000
      <c .narration>To <i>WebVTT</i>.
    
  • Comments:
    Both, comments within cues and complete cues commented out are possible, e.g.

    WEBVTT
    
     COMMENT -->
     00:02.000 --> 00:03.000
     two; this is entirely
     commented out
     
     00:06.000 --> 00:07.000
     this part of the cue is visible
     <! this part isn't >
     <and neither is this>
    

Finally, I believe we still need to add the following features:

  • Language tags:
    I’d like to add a language tag that allows to mark up a subpart of cue text as being in a different language. We need this feature for mixed-language cues (in particular where a different font may be necessary for the inline foreign-language text). But more importantly we will need this feature for cues that contain text descriptions rather than captions, such that a speech synthesizer can pick the correct language model to speak the foreign-language text. It was discussed that this could be done with a <lang jp>xxx</lang> type of markup.
  • Roll-up captions:
    When we use timestamp objects and the future text is hidden, then is un-hidden upon reaching its time, we should allow the cue text to scroll up a line when the un-hidden text requires adding a new line. This is the typical way in which TV live captions have been displayed and so users are acquainted with this display style.
  • Inline navigation:
    For chapter tracks the primary use of cues are for navigation. In other formats – in particular in DAISY-books for blind users – there are hierarchical navigation possibilities within media resources. We can use timestamp objects to provide further markers for navigation within cues, but in order to make these available in a hierarchical fashion, we will need a grouping tag. It would be possible to introduce a <nav> tag that can group several timestamp objects for navigation.
  • Default caption width:
    At the moment, the default display size of a caption cue is 100% of the video’s width (height for vertical directions), which can be overruled with the “S” cue setting. I think it should by default rather be the width (height) of the bounding box around all the text inside the cue.

Aside from these changes to WebVTT, there are also some things that can be improved on the <track> element. I personally support the introduction of the source element underneath the track element, because that allows us to provide different caption files for different devices through the @media media queries attribute and it allows support for more than just one default captioning format. This change needs to be made soon so we don’t run into trouble with the currently empty track element.

I further think a oncuelistchange event would be nice as well in cases where the number of tracks is somehow changed – in particular when coming from within a media file.

Other than this, I’m really very happy with the state that we have achieved this far.

HTML5 Video Presentations at LCA 2011

Working in the WHAT WG and the W3C HTML WG, you sometimes forget that all the things that are being discussed so heatedly for standardization are actually leading to some really exciting new technologies that not many outside have really taken note of yet.

This week, during the Australian Linux Conference in Brisbane, I’ve been extremely lucky to be able to show off some awesome new features that browser vendors have implemented for the audio and video elements. The feedback that I got from people was uniformly plain surprise – nobody expected browser to have all these capabilities.

The examples that I showed off have mostly been the result of working on a book for almost 9 months of the past year and writing lots of examples of what can be achieved with existing implementations and specifications. They have been inspired by diverse demos that people made in the last years, so the book is linking to many more and many more amazing demos.

Incidentally, I promised to give a copy of the book away to the person with the best idea for a new Web application using HTML5 media. Since we ran out of time, please shoot me an email or a tweet (@silviapfeiffer) within the next 4 weeks and I will send another copy to the person with the best idea. The copy that I brought along was given to a student who wanted to use HTML5 video to display on surfaces of 3D moving objects.

So, let’s get to the talks.

On Monday, I gave a presentation on “Audio and Video processing in HTML5“, which had a strong focus on the Mozilla Audio API.

I further gave a brief lightning talk about “HTML5 Media Accessibility Update“. I am expecting lots to happen on this topic during this year.

Finally, I gave a presentation today on “The Latest and Coolest in HTML5 Media” with a strong focus on video, but also touching on audio and media accessibility.

The talks were streamed live – congrats to Ryan Verner for getting this working with support from Ben Hutchings from DebConf and the rest of the video team. The videos will apparently be available from http://linuxconfau.blip.tv/ in the near future.

UPDATE 4th Feb 2011: And here is my LCA talk …

with subtitles on YouTube:

adaptive HTTP streaming for open codecs

At this week’s FOMS in New York we had one over-arching topic that seemed to be of interest to every single participant: how to do adaptive bitrate streaming over HTTP for open codecs. On the first day, there was a general discussion about the advantages and disadvantages of adaptive HTTP streaming, while on the second day, we moved towards designing a solution for Ogg and WebM. While I didn’t attend all the discussions, I want to summarize the insights that I took out of the days in this blog post and the alternative implementation strategies that were came up with.

Use Cases for Adaptive HTTP Streaming

Streaming using RTP/RTSP has in the past been the main protocol to provide live video streams, either for broadcast or for real-time communication. It has been purpose-built for chunked video delivery and has features that many customers want, such as the ability to encrypt the stream, to tell players not to store the data, and to monitor the performance of the stream such that its bandwidth can be adapted. It has, however, also many disadvantages, not least that it goes over ports that normal firewalls block and thus is rather difficult to deploy, but also that it requires special server software, a client that speaks the protocol, and has a signalling overhead on the transport layer for adapting the stream.

RTP/RTSP has been invented to allow for high quality of service video consumption. In the last 10 years, however, it has become the norm to consume “canned” video (i.e. non-live video) over HTTP, making use of the byte-range request functionality of HTTP for seeking. While methods have been created to estimate the size of a pre-buffer before starting to play back in order to achieve continuous playback based on the bandwidth of your pipe at the beginning of downloading, not much can be done when one runs out of pre-buffer in the middle of playback or when the CPU on the machine doesn’t manage to catch up with decoding of the sheer amount of video data: your playback stops to go into re-buffering in the first case and starts to become choppy in the latter case.

An obvious approach to improving this situation is the scale the bandwidth of the video stream down, potentially even switch to a lower resolution video, right in the middle of playback. Apple’s HTTP live streaming, Microsoft’s Smooth Streaming, and Adobe’s Dynamic Streaming are all solutions in this space. Also, ISO/MPEG is working on DASH (Dynamic Adaptive Streaming over HTTP) is an effort to standardize the approach for MPEG media. No solution yets exist for the open formats within Ogg or WebM containers.

Some features of HTTP adaptive streaming are:

  • Enables adaptation of downloading to avoid continuing buffering when network or machine cannot cope.
  • Gapless switching between streams of different bitrate.
  • No special server software is required – any existing Web Server can be used to provide the streams.
  • The adaptation comes from the media player that actually knows what quality the user experiences rather than the network layer that knows nothing about the performance of the computer, and can only tell about the performance of the network.
  • Adaptation means that several versions of different bandwidth are made available on the server and the client switches between them based on knowledge it has about the video quality that the user experiences.
  • Bandwidth is not wasted by downloading video data that is not being consumed by the user, but rather content is pulled moments just before it is required, which works both for the live and canned content case and is particularly useful for long-form content.

Viability

In discussions at FOMS it was determined that mid-stream switching between different bitrate encoded audio files is possible. Just looking at the PCM domain, it requires stitching the waveform together at the switch-over point, but that is not a complex function. To be able to do that stitching with Vorbis-encoded files, there is no need for a overlap of data, because the encoded samples of the previous window in a different bitrate page can be used as input into the decoding of the current bitrate page, as long as the resulting PCM samples are stitched.

For video, mid-stream switching to a different bitrate encoded stream is also acceptable, as long as the switch-over point adheres to a keyframe, which can be independently decoded.

Thus, the preparation of the alternative bitstream videos requires temporal synchronisation of keyframes on video – the audio can deal with the switch-over at any point. A bit of intelligent encoding is thus necessary – requiring the encoding pipeline to provide regular keyframes at a certain rate would be sufficient. Then, the switch-over points are the keyframes.

Technical Realisation

With the solutions from Adobe, Microsoft and Apple, the technology has been created such there are special tools on the server that prepare the content for adaptive HTTP streaming and provide a manifest of the prepared content. Typically, the content is encoded in versions of different bitrates and the bandwidth versions are broken into chunks that can be decoded independently. These chunks are synchronised between the different bitrate versions such that there are defined switch-over points. The switch-over points as well as the file names of the different chunks are documented inside a manifest file. It is this manifest file that the player downloads instead of the resource at the beginning of streaming. This manifest file informs the player of the available resources and enables it to orchestrate the correct URL requests to the server as it progresses through the resource.

At FOMS, we took a step back from this approach and analysed what the general possibilities are for solving adaptive HTTP streaming. For example, it would be possible to not chunk the original media data, but instead perform range requests on the different bitrate versions of the resource. The following options were identified.

Chunking

With Chunking, the original bitrate versions are chunked into smaller full resources with defined switch-over points. This implies creation of a header on each one of the chunks and thus introduces overhead. Assuming we use 10sec chunks and 6kBytes per chunk, that results in 5kBit/sec extra overhead. After chunking the files this way, we provide a manifest file (similar to Apple’s m3u8 file, or the SMIL-based manifest file of Microsoft, or Adobe’s Flash Media Manifest file). The manifest file informs the client about the chunks and the switch-over points and the client requests those different resources at the switch-over points.

Disadvantages:

  • Header overhead on the pipe.
  • Switch-over delay for decoding the header.
  • Possible problem with TCP slowstart on new files.
  • A piece of software is necessary on server to prepare the chunked files.
  • A large amount of files to manage on the server.
  • The client has to hide the switching between full resources.

Advantages:

  • Works for live streams, where increasing amounts of chunks are written.
  • Works well with CDNs, because mid-stream switching to another server is easy.
  • Chunks can be encoded such that there is no overlap in the data necessary on switch-over.
  • May work well with Web sockets.
  • Follows the way in which proprietary solutions are doing it, so may be easy to adopt.
  • If the chunks are concatenated on the client, you get chained Ogg files (similar concept in WebM?), which are planned to be supported by Web browsers and are thus legal files.

Chained Chunks

Alternatively to creating the large number of files, one could also just create the chained files. Then, the switch-over is not between different files, but between different byte ranges. The headers still have to be read and parsed. And a manifest file still has to exist, but it now points to byte ranges rather than different resources.

Advantages over Chunking:

  • No TCP-slowstart problem.
  • No large number of files on the server.

Disadvantages over Chunking:

  • Mid-stream switching to other servers is not easily possible – CDNs won’t like it.
  • Doesn’t work with Web sockets as easily.
  • New approach that vendors will have to grapple with.

Virtual Chunks

Since in Chained Chunks we are already doing byte-range requests, it is a short step towards simply dropping the repeating headers and just downloading them once at the beginning for all possible bitrate files. Then, as we seek to different positions in “the” file, the byte range of the bitrate version that makes sense to retrieve at that stage would be requested. This could even be done with media fragment URIs, through addressing with time ranges is less accurate than explicit byte ranges.

In contrast to the previous two options, this basically requires keeping n different encoding pipelines alive – one for every bitrate version. Then, the byte ranges of the chunks will be interpreted by the appropriate pipeline. The manifest now points to keyframes as switch-over points.

Advantage over Chained Chunking:

  • No header overhead.
  • No continuous re-initialisation of decoding pipelines.

Disadvantages over Chained Chunking:

  • Multiple decoding pipelines need to be maintained and byte ranges managed for each.

Unchunked Byte Ranges

We can even consider going all the way and not preparing the alternative bitrate resources for switching, i.e. not making sure that the keyframes align. This will then require the player to do the switching itself, determine when the next keyframe comes up in its current stream then seek to that position in the next stream, always making sure to go back to the last keyframe before that position and discard all data until it arrives at the same offset.

Disadvantages:

  • There will be an overlap in the timeline for download, which has to be managed from the buffering and alignment POV.
  • Overlap poses a challenge of downloading more data than necessary at exactly the time where one doesn’t have bandwidth to spare.
  • Requires seeking.
  • Messy.

Advantages:

  • No special authoring of resources on the server is needed.
  • Requires a very simple manifest file only with a list of alternative bitrate files.

Final concerns

At FOMS we weren’t able to make a final decision on how to achieve adaptive HTTP streaming for open codecs. Most agreed that moving forward with the first case would be the right thing to do, but the sheer number of files that can create is daunting and it would be nice to avoid that for users.

Other goals are to make it work in stand-alone players, which means they will need to support loading the manifest file. And finally we want to enable experimentation in the browser through JavaScript implementation, which means there needs to be an interface to provide the quality of decoding to JavaScript. Fortunately, a proposal for such a statistics API already exists. The number of received frames, the number of dropped frames, and the size of the video are the most important statistics required.

VP8/WebM: Adobe is the key to open video on the Web

Google have today announced the open sourcing of VP8 and the creation of a new media format WebM.

Technical Challenges

As I predicted earlier, Google had to match VP8 with an audio codec and a container format – their choice was a subpart of the Matroska format and the Vorbis codec. To complete the technical toolset, Google have:

  • developed ffmpeg patches, so an open source encoding tool for WebM will be available
  • developed GStreamer and DirectShow plugins, so players that build on these frameworks will be able to decode WebM,
  • and developed an SDK such that commercial partners can implement support for WebM in their products.

This has already been successful and several commercial software products are already providing support for WebM.

Google haven’t forgotten the mobile space either – a bunch of Hardware providers are listed as supporters on the WebM site and it can be expected that developments have started.

The speed of development of software and hardware around WebM is amazing. Google have done an amazing job at making sure the technology matures quickly – both through their own developments and by getting a substantial number of partners included. That’s just the advantage of being Google rather than a Xiph, but still an amazing achievement.

Browsers

As was to be expected, Google managed to get all the browser vendors that are keen to support open video to also support WebM: Chrome, Firefox and Opera all have come out with special builds today that support WebM. Nice work!

What is more interesting, though, is that Microsoft actually announced that they will support WebM in future builds of IE9 – not out of the box, but on systems where the codec is already installed. Technically, that is be the same situation as it will be for Theora, but the difference in tone is amazing: in this blog post, any codec apart from H.264 was condemned and rejected, but the blog post about WebM is rather positive. It signals that Microsoft recognize the patent risk, but don’t want to be perceived of standing in the way of WebM’s uptake.

Apple have not yet made an announcement, but since it is not on the list of supporters and since all their devices exclusively support H.264 it stands to expect that they will not be keen to pick up WebM.

Publishers

What is also amazing is that Google have already achieved support for WebM by several content providers. The first of these is, naturally, YouTube, which is offering a subset of its collection also in the WebM format and they are continuing to transcode their whole collection. Google also has Brightcov, Ooyala, and Kaltura on their list of supporters, so content will emerge rapidly.

Uptake

So, where do we stand with respect to a open video format on the Web that could even become the baseline codec format for HTML5? It’s all about uptake – if a substantial enough ecosystem supports WebM, it has all chances of becoming a baseline codec format – and that would be a good thing for the Web.

And this is exactly where I have the most respect for Google. The main challenge in getting uptake is in getting the codec into the hands of all people on the Internet. This, in particular, includes people working on Windows with IE, which is still the largest browser from a market share point of view. Since Google could not realistically expect Microsoft to implement WebM support into IE9 natively, they have found a much better partner that will be able to make it happen – and not just on Windows, but on many platforms.

Yes, I believe Adobe is the key to creating uptake for WebM – and this is admittedly something I have completely overlooked previously. Adobe has its Flash plugin installed on more than 90% of all browsers. Most of their users will upgrade to a new version very soon after it is released. And since Adobe Flash is still the de-facto standard in the market, it can roll out a new Flash plugin version that will bring WebM codec support to many many machines – in particular to Windows machines, which will in turn enable all IE9 users to use WebM.

Why would Adobe do this and thus cement its Flash plugin’s replacement for video use by HTML5 video? It does indeed sound ironic that the current market leader in online video technology will be the key to creating an open alternative. But it makes a lot of sense to Adobe if you think about it.

Adobe has itself no substantial standing in codec technology and has traditionally always had to license codecs. Adobe will be keen to move to a free codec of sufficient quality to replace H.264. Also, Adobe doesn’t earn anything from the Flash plugins themselves – their source of income are their authoring tools. All they will need to do to succeed in a HTML5 WebM video world is implement support for WebM and HTML5 video publishing in their tools. They will continue to be the best tools for authoring rich internet applications, even if these applications are now published in a different format.

Finally, in the current hostile space between Apple and Adobe related to the refusal of Apple to allow Flash onto its devices, this may be the most genius means of Adobe at getting back at them. Right now, it looks as though the only company that will be left standing on the H.264-only front and outside the open WebM community will be Apple. Maybe implementing support for Theora wouldn’t have been such a bad alternative for Apple. But now we are getting a new open video format and it will be of better quality and supported on hardware. This is exciting.

IP situation

I cannot, however, finish this blog post on a positive note alone. After reading the review of VP8 by a x.264 developer, it seems possible that VP8 is infringing on patents that are outside the patent collection that Google has built up in codecs. Maybe Google have calculated with the possibility of a patent suit and put money away for it, but Google certainly haven’t provided indemnification to everyone else out there. It is a tribute to Google’s achievement that given a perceived patent threat – which has been the main inhibitor of uptake of Theora – they have achieved such an uptake and industry support around VP8. Hopefully their patent analysis is sound and VP8 is indeed a safe choice.

UPDATE (22nd May): After having thought about patents and the situation for VP8 a bit more, I believe the threat is really minimal. You should also read these thoughts of a Gnome developer, these of a Debian developer and the emails on the Theora mailing list.

Introducing media accessibility into HTML5

In recent months, people in the W3C HTML5 Accessibility Task Force developed two proposals for introducing caption, subtitle, and more generally time-aligned text support into HTML5 audio and video.

These time-aligned text files can either come as external files that are associated with the timeline of the media resource, or they come as part of the media resource in a binary track.

For both cases we now have proposals to extend the HTML5 specification.

Firstly, let’s look at time-aligned text in external files. The change proposal introduces markup to associate such external files as a kind of “virtual track” with a media resource. Here is an example:

<video src="video.ogv">
<track src="video_cc.ttml" type="application/ttaf+xml" language="en" role="caption"></track>
<track src="video_tad.srt" type="text/srt" language="en" role="textaudesc"></track>
<trackgroup role="subtitle">
<track src="video_sub_en.srt" type="text/srt; charset='Windows-1252'" language="en"></track>
<track src="video_sub_de.srt" type="text/srt; charset='ISO-8859-1'" language="de"></track>
<track src="video_sub_ja.srt" type="text/srt; charset='EUC-JP'" language="ja"></track>
</trackgroup>
</video>

The video resource is “video.ogv”. Associated with it are five timed text resources.

The first one is written in TTML (which is the new name for DFXP), is a caption track and in English. TTML is particularly useful when you want to provide more than just an unformatted piece of text to the viewers. Hearing-impaired users appreciate any visual help they can be provided with to absorb the caption text more quickly. This includes colour coding of speakers, positioning of text close to the speaking person on screen, or even animated musical notes to signify music. Thus, a format like TTML that allows for formatting and positioning information is an appropriate format to specify captions.

All other timed text resources are provided in SRT format, which is a simpler format that TTML with only plain text in the text cues.

The second text track is a textual audio description track. A textual audio description is in fact targeted at the vision-impaired and contains text that is expected to be read out by a screen reader or routed to a braille device. Thus, as the video plays, a vision-impaired user receives additional information about the visual content of the scene through their screen reader or braille device. The SRT format is particularly useful for providing textual audio descriptions since it only provides plain text, which can easily be handed on to assistive technology. When authoring such textual audio descriptions, it is very important to pick time intervals in the original media resource where no other significant audio cue is provided, such that the vision-impaired user is able to listen to the screen reader during that time.

The last three text tracks are subtitle tracks. They are grouped into a trackgroup element, which is not strictly necessary, but enables the author to say that these tracks are supposed to be alternatives. Thus, a Web Browser can create a menu with all the available tracks and put the tracks in the trackgroup into a menu of their own where only one option is selectable (similar to how radiobuttons work). Incidentally, the trackgroup element also allows to avoid having to repeat the role attribute in all the containing tracks. It is expected that these menus will be added to the default media controls and will thus be visible if the media element has a controls attribute.

With the role, type and language attributes, it is easy for a Web Browser to understand what the different tracks have to offer. A Web Browser can even decide to offer new functionality that is helpful to certain user groups. For example, an addition to a Web Browser’s default settings could be to allow users to instruct a Web Browser to always turn on captions or subtitles if they are available in the user’s main language. Or to always turn on textual audio descriptions. In this way, a user can customise their default experience of a media resource over and on top of what a Web page author decides to expose.

Incidentally, the choice of “track” as a name for relating external text resources to a media element has a deeper meaning. It is easily possible in future to extend “track” elements to not just point to dependent text resources, but also to dependent audio or video resources. For example, an actual audio description that is a recording of a human voice rather than a rendered text description could be association in the same way. Right now, such an implementation is not envisaged by the Browser vendors, but it will be something to work towards in the future.

Now, with such functionality available, there is naturally a desire to be able to control activation or de-activation of text tracks through JavaScript, not just through user interaction. A Web Developer may for example want to override the default controls provided by a Web Browser and run their own JavaScript-based controls, thus requiring to create their own selection menu for the tracks.

This is actually also an issue more generally and applies to all track types, including such tracks that come inside an existing media resource. In the current specification such tracks are not exposed and can therefore not be activated.

This is where the second specification that the W3C Accessibility Task Force has worked towards comes in: the media multitrack JavaScript API.

This specification introduces a read-only JavaScript interface to the audio and video elements to allow Web Developers to find out about the tracks (including the virtual tracks) that a media resource offers. The only action that the interface currently provides is to enable or disable tracks.
Here is an example use to turn on a french subtitle track:

if (video.tracks[2].role == "subtitle" && video.tracks[2].language == "fr") video.tracks[2].enabled = true;

There is still a need to introduce a means to actually expose the text cues as they relate to the currentTime of the media resource. This has not yet been specified in the given proposals.

The text cues could be exposed in several ways. They could be exposed through introducing an event, i.e. every time a new text cue becomes active, a callback is called which is given the active text cue (if such a callback had been registered previously). Another option is to simply write the text cues into a specified div-element in the DOM and thus expose them directly in the Browser. A third idea could be to expose the text cues in an iframe-like element to avoid any cross-site security issues. And a fourth idea that we have discussed is to expose the text cues in an attribute of the track.

All of this obviously also relates to how to actually render the text cues and whether to render them in a shadow DOM so as to make the JavaScript reading separate from the rendering and address security and copyright issues. I’d be curious in opinions here on how it should be done.

How to display seeked position for HTML5 video

Recently, I was asked for some help on coding with an HTML5 video element and its events. In particular the question was: how do I display the time position that somebody seeked to in a video?

Here is a code snipped that shows how to use the seeked event:


<video onseeked="writeVideoTime(this.currentTime);" src="video.ogv" controls></video>
<p>position:</p><div id="videotime"></div>
<script type="text/javascript">
// get video element
var video = document.getElementsByTagName("video")[0];
function writeVideoTime(t) {
document.getElementById("videotime").innerHTML=t;
}
</script>

Other events that can be used in a similar way are:

  • loadstart: UA requests the media data from the server
  • progress: UA is fetching media data from the server
  • suspend: UA is on purpose idling on the server connection mid-fetching
  • abort: UA aborts fetching media data from the server
  • error: UA aborts fetching media because of a network error
  • emptied: UA runs out of network buffered media data (I think)
  • stalled: UA is waiting for media data from the server
  • play: playback has begun after play() method returns
  • pause: playback has been paused after pause() method returns
  • loadedmetadata: UA has received all its setup information for the media resource, duration and dimensions and is ready to play
  • loadeddata: UA can render the media data at the current playback position for the first time
  • waiting: playback has stopped because the next frame is not available yet
  • playing: playback has started
  • canplay: playback can resume, but at risk of buffer underrun
  • canplaythrough: playback can resume without estimated risk of buffer underrun
  • seeking: seeking attribute changed to true (may be too short to catch)
  • seeked: seeking attribute changed to false
  • timeupdate: current playback position changed enough to report on it
  • ended: playback stopped at media resource end; ended attribute is true
  • ratechange: defaultPlaybackRate or playbackRate attribute have just changed
  • durationchange: duration attribute has changed
  • volumechange:volume attribute or the muted attribute has changed

Please refer to the actual event list in the specification for more details and more accurate information on the events.

Audio Track Accessibility for HTML5

I have talked a lot about synchronising multiple tracks of audio and video content recently. The reason was mainly that I foresee a need for more than two parallel audio and video tracks, such as audio descriptions for the vision-impaired or dub tracks for internationalisation, as well as sign language tracks for the hard-of-hearing.

It is almost impossible to introduce a good scheme to deliver the right video composition to a target audience. Common people will prefer bare a/v, vision-impaired would probably prefer only audio plus audio descriptions (but will probably take the video), and the hard-of-hearing will prefer video plus captions and possibly a sign language track . While it is possible to dynamically create files that contain such tracks on a server and then deliver the right composition, implementation of such a server method has not been very successful in the last years and it would likely take many years to roll out such new infrastructure.

So, the only other option we have is to synchronise completely separate media resource together as they are selected by the audience.

It is this need that this HTML5 accessibility demo is about: Check out the demo of multiple media resource synchronisation.

I created a Ogg video with only a video track (10m53s750). Then I created an audio track that is the original English audio track (10m53s696). Then I used a Spanish dub track that I found through BlenderNation as an alternative audio track (10m58s337). Lastly, I created an audio description track in the original language (10m53s706). This creates a video track with three optional audio tracks.

I took away all native controls from these elements when using the HTML5 audio and video tag and ran my own stop/play and seeking approaches, which handled all media elements in one go.

I was mostly interested in the quality of this experience. Would the different media files stay mostly in sync? They are normally decoded in different threads, so how big would the drift be?

The resulting page is the basis for such experiments with synchronisation.

The page prints the current playback position in all of the media files at a constant interval of 500ms. Note that when you pause and then play again, I am re-synching the audio tracks with the video track, but not when you just let the files play through.

I have let the files play through on my rather busy Macbook and have achieved the following interesting drift over the course of about 9 minutes:

Drift between multiple parallel played media elements

You will see that the video was the slowest, only doing roughly 540s, while the Spanish dub did 560s in the same time.

To fix such drifts, you can always include regular re-synchronisation points into the video playback. For example, you could set a timeout on the playback to re-sync every 500ms. Within such a short time, it is almost impossible to notice a drift. Don’t re-load the video, because it will lead to visual artifacts. But do use the video’s currentTime to re-set the others. (UPDATE: Actually, it depends on your situation, which track is the best choice as the main timeline. See also comments below.)

It is a workable way of associating random numbers of media tracks with videos, in particular in situations where the creation of merged files cannot easily be included in a workflow.