Cloud Text-To-Speech #

Single interface to Google, Microsoft, and Amazon Text-To-Speech. Flutter implementation of:

• Google Cloud Text-To-Speech API
• Microsoft Azure Cognitive Text-To-Speech API
• Amazon Polly API

Features #

• Universal implementation for accessing all providers with one interface.
• Separate implementation for every provider so we could access every functionality.
• Sanitize SSML input per provider so we send only supported SSML elements.
• Locale names in English and native language so we could display language selector.
• Fake name generation for Google voices that are generated randomly based on voice locale.
• Accessible configurable output format (per provider), rate, and pitch.

Getting Started #

There are essentially two ways to use Cloud Text-To-Speech:

• Universal: Using TtsUniversal to be able to configure the TTS provider dynamically and us it.
  • Single: Using Providers.google, Providers.microsoft, Providers.amazon to use the single provider at a time.
  • Combine: Using Providers.combine to combine all providers and get all voices at once.
• Provider: Using TtsGoogle, TtsMicrosoft, TtsAmazon to get the most from provider's API.

Universal(Single) #

To init configuration use:

    //Do init once and run it before any other method
    TtsUniversal.init(
        provider: Providers.amazon,
        googleParams: InitParamsGoogle(apiKey: 'API-KEY'),
        microsoftParams: InitParamsMicrosoft(
        subscriptionKey: 'SUBSCRIPTION-KEY', region: 'eastus'),
        amazonParams: InitParamsAmazon(
        keyId: 'KEY-ID', accessKey: 'ACCESS-KEY', region: 'us-east-1'),
        withLogs: true
    );

To change provider use:

    TtsUniversal.setProvider(Providers.microsoft);

To get the list of all voices use:

    // Get voices
    final voicesResponse = await TtsUniversal.getVoices();
    final voices = voicesResponse.voices;

    //Print all available voices
    print(voices);

    //Pick an English Voice
    final voice = voices
      .where((element) => element.locale.code.startsWith("en-"))
      .toList(growable: false)
      .first;

To convert TTS and get audio use:

    //Generate Audio for a text
    const text = "Amazon, Microsoft and Google Text-to-Speech API are awesome";

    final ttsParams = TtsParamsUniversal(
        voice: voice,
        audioFormat: AudioOutputFormatUniversal.mp3_64k,
        text: text,
        rate: 'slow', // optional
        pitch: 'default' // optional
    );
    
    final ttsResponse = await TtsUniversal.convertTts(ttsParams);
    
    //Get the audio bytes.
    final audioBytes = ttsResponse.audio.buffer.asByteData();

Universal(Combine) #

To init configuration use:

    //Do init once and run it before any other method
    TtsUniversal.init(
        provider: Providers.combine,
        googleParams: InitParamsGoogle(apiKey: 'API-KEY'),
        microsoftParams: InitParamsMicrosoft(
        subscriptionKey: 'SUBSCRIPTION-KEY', region: 'eastus'),
        amazonParams: InitParamsAmazon(
        keyId: 'KEY-ID', accessKey: 'ACCESS-KEY', region: 'us-east-1'),
        withLogs: true
    );

To change provider use:

    TtsUniversal.setProvider(Providers.combine);

To get the list of all voices use:

    // Get voices
    final voicesResponse = await TtsUniversal.getVoices();
    final voices = voicesResponse.voices;

    //Print all available voices
    print(voices);

    //Pick an English Voice
    final voice = voices
      .where((element) => element.locale.code.startsWith("en-"))
      .toList(growable: false)
      .first;

To convert TTS and get audio use:

    //Generate Audio for a text
    const text = "Amazon, Microsoft and Google Text-to-Speech API are awesome";

    final ttsParams = TtsParamsUniversal(
        voice: voice,
        audioFormat: AudioOutputFormatUniversal.mp3_64k,
        text: text,
        rate: 'slow', // optional
        pitch: 'default' // optional
    );
    
    final ttsResponse = await TtsUniversal.convertTts(ttsParams);
    
    //Get the audio bytes.
    final audioBytes = ttsResponse.audio.buffer.asByteData();

Google #

To init configuration use:

    //Do init once and run it before any other method
    TtsGoogle.init(params: InitParamsGoogle(apiKey: "API-KEY"), withLogs: true);

To get the list of all voices use:

    // Get voices
    final voicesResponse = await TtsGoogle.getVoices();
    final voices = voicesResponse.voices;

    //Print all voices
    print(voices);

    //Pick an English Voice
    final voice = voices
        .where((element) => element.locale.code.startsWith("en-"))
        .toList(growable: false)
        .first;

To convert TTS and get audio use:

   //Generate Audio for a text
  final text = '<speak>Google<break time="2s"> Speech Service Text-to-Speech API is awesome!</speak>';

  TtsParamsGoogle ttsParams = TtsParamsGoogle(
      voice: voice,
      audioFormat: AudioOutputFormatGoogle.mp3,
      text: text,
      rate: 'slow', // optional
      pitch: 'default' // optional
  );

  final ttsResponse = await TtsGoogle.convertTts(ttsParams);

  //Get the audio bytes.
  final audioBytes = ttsResponse.audio.buffer.asByteData();

Microsoft #

To init configuration use:

    //Do init once and run it before any other method
    TtsMicrosoft.init(
        params: InitParamsMicrosoft(
        subscriptionKey: "SUBSCRIPTION-KEY", region: "eastus"),
        withLogs: true
    );

To get the list of all voices use:

    // Get voices
    final voicesResponse = await TtsMicrosoft.getVoices();
    final voices = voicesResponse.voices;

    //Print all voices
    print(voices);

    //Pick an English Voice
    final voice = voices
        .where((element) => element.locale.code.startsWith("en-"))
        .toList(growable: false)
        .first;

To convert TTS and get audio use:

   //Generate Audio for a text
  final text = '<speak>Microsoft<break time="2s"> Speech Service Text-to-Speech API is awesome!</speak>';

  TtsParamsMicrosoft ttsParams = TtsParamsMicrosoft(
      voice: voice,
      audioFormat: AudioOutputFormatMicrosoft.audio48Khz192kBitrateMonoMp3,
      text: text,
      rate: 'slow', // optional
      pitch: 'default' // optional
  );

  final ttsResponse = await TtsMicrosoft.convertTts(ttsParams);

  //Get the audio bytes.
  final audioBytes = ttsResponse.audio.buffer.asByteData();

Amazon #

To init configuration use:

    //Do init once and run it before any other method
    TtsAmazon.init(
        params: InitParamsAmazon(keyId: 'KEY-ID', accessKey: 'ACCESS-KEY', region: 'us-east-1'),
        withLogs: true
    );

To get the list of all voices use:

    // Get voices
    final voicesResponse = await TtsAmazon.getVoices();
    final voices = voicesResponse.voices;

    //Print all voices
    print(voices);

    //Pick an English Voice
    final voice = voices
        .where((element) => element.locale.code.startsWith("en-"))
        .toList(growable: false)
        .first;

To convert TTS and get audio use:

   //Generate Audio for a text
  final text = '<speak>Amazon<break time="2s"> Speech Service Text-to-Speech API is awesome!</speak>';

  TtsParamsAmazon ttsParams = TtsParamsAmazon(
      voice: voice,
      audioFormat: AudioOutputFormatAmazon.audio48Khz192kBitrateMonoMp3,
      text: text,
      rate: 'slow', // optional
      pitch: 'default' // optional
  );

  final ttsResponse = await TtsAmazon.convertTts(ttsParams);

  //Get the audio bytes.
  final audioBytes = ttsResponse.audio.buffer.asByteData();

Notes #

There are things you should take care of:

• Securing of your API keys and credentials, they could be extracted from your mobile app.
• Sometimes Amazon Polly is not working in emulator, so you could get 403 error.
• For fixing SSML/XML before passing it to TTS Params, you could use the xml packages, method XmlDocument.parse(ssml).toXmlString().
• Audio has uniform format for all providers, it is Uint8List that you could use to play it or save it to file.
• Some player packages that are good fit are: audioplayers and assets_audio_player.