Self Voicing panel/enactor for UIOptions.

[Colin Clark]

Just as a further update to this, Web Audio support has been delayed by Mozilla until October due to politics in the working group regarding the spec.

This confirms to me that we should look at normalization libraries for the HTML5 <audio> element, rather than Web Audio until it stabilizes in the fall.

Colin

---
Colin Clark
http://fluidproject.org

On 2013-07-26, at 9:56 AM, "Zenevich, Yura" <yzenevich at ocadu.ca> wrote:

> That's a good idea, Colin.
> 
> I will take a look at that.
> 
> Yura
> 
> On 2013-07-26, at 9:50 AM, Colin Clark <colinbdclark at gmail.com> wrote:
> 
>> Hi Yura,
>> 
>> If we're really stuck, I think we should take a look at loading audio files into AudioBuffers and playing them back with the WebAudio API using an AudioBufferSourceNode. It looks like Firefox will have full support for this shipping in the next release, which is only a few weeks away.
>> 
>> I can whip up some code to do this next week if it is helpful.
>> 
>> Colin
>> 
>> ---
>> Colin Clark
>> http://fluidproject.org
>> 
>> On 2013-07-26, at 12:10 AM, "Zenevich, Yura" <yzenevich at ocadu.ca> wrote:
>> 
>>> Hi everyone,
>>> 
>>> Just a quick update on the self voicing work I've been doing in the past couple of days.
>>> 
>>> I've made some improvements to the existing version of the self voicing enactor. For example, there's no more limitation (introduced due to the Google Translate TTS) of 100 characters max at a time. Right now the text to speech engine will utter a complete text node at a time. There's also a better handling of media events which eliminated the pre-existing timeout hacks in between playback and errors.
>>> 
>>> The not so good news is that the self voicing enactor is almost non-working in Firefox. It seems that the the audio element events just die shortly after the playback has started. This happens due to the fact that the browser does not seem to properly handle the update of the audio's src attribute:
>>> 
>>> Expected behaviour would be an |canplaythrough| event being fired (which would trigger the enactor).
>>> 
>>> Actual behaviour is an incorrect handling of the response to the TTS engine server that results in the following response message:
>>> Reload the page to get source for: (url to tts engine server)
>>> I've also experimented with replacing the audio tag completely and with not attaching the audio tag to the document itself but to no success.
>>> 
>>> Perhaps some of you have suggestions for different scenarios that could be tested.
>>> 
>>> Regards,
>>> 
>>> Yura
>> 
>