• Blog
  • Archives
  • Code
  • Talks
  • About

Updated Google text-to-speech library and added to Bower

Note: This post is over 2 years old and may now be out of date. Some of the links not work anymore.

(2 minute read)

Just pushed a few updates to the Google text-to-speech API library - google-tts - I first put out a while back. Here are the higlights...

Large text gets split up

It turns out that in Google Translate if you ask it to read out more than 100 characters in one go it actually splits up the text into 100 character slices, making consecutive audio requests for each slice. Thanks to Julien Synx for pointing this out. Google-tts now does this too.

Callback when playback is finished

Previously when calling play() you callback would be triggered once the song had started playing. This now gets triggered once the song has finished playing. I was holding off on implementing this as I wasn't sure if it was possible to detect when HTML5 audio playback had completed in an easy cross-browser compatible away. Now I've decided to just go with the ended event as I think we've reached the point where all decent browsers support it.

Pluggable playback mechanisms

The current version of Firefox (21) does not support the playback of MP3s using HTML5 audio, though I'm told that soon-to-be released versions might. In any case, in order to provide a solution for Firefox right now I opted to incorporate the excellent SoundManager2 library into this project (yes unfortunately, this is a Flash-based solution for Firefox). I've tried to de-couple this library from GoogleTTS as much as possible, especially since not everybody will need or even want to use SoundManager2.

There is now a GoogleTTS.Player base class (should be thought of more as an interface) from which different playback mechanisms must inherit. It looks like this:

* Represents a playback mechanism.
* @constructor
TTS.Player = function() {
  var self = this;

  * Get whether this playback mechanism is available for use.
  * @param cb Function Result callback (Boolean available)
  self.available = function(cb) { throw new Error('Not yet implemented'); };

  * Play given URL.
  * @param url String
  * @param cb Function Called after we finish playing (Error err)
  self.play = function(url, cb) { throw new Error('Not yet implemented'); };

  * Get name of this player.
  self.toString = function() { throw new Error('Not yet implemented'); };

Two built-in playback mechanisms are provided within the library - HTML5 audio and SoundManager2. You can easily define, instantiate and add your own to the library using addPlayer() of a GoogleTTS instance. The library will use whichever playback mechanism is first found to be 'available', i.e. usable in the current browser context so the order in which you register playback mechanisms matters.

Testing -> Bower

We now have some basic unit tests...yay. Continuous build has also been setup - https://travis-ci.org/hiddentao/google-tts.

I've refrained from adding google-tts to NPM as it's not really usable on the back-end. But I've now added it to Bower, so hopefully more people will become aware of its existence.