Synthesize

Text-To-Speach Service: Synthesize

There two ways to synthesize speech from text using Dasha tts service: the TTS API (particularly, the synthesize function) and the CLI tts synthesize command:

Usage: dasha tts synthesize|synth [options] <text>

Options:
  -o --output <filename>
  --provider-name <name>  TTS provider name (default: "dasha")
  --lang <code>            (default: "en-US")
  --speaker <name>         (default: "default")
  --emotion <emotion>      (default: "neutral")
  --speed <value>          (default: 1)
  --variation <value>      (default: 0)

They have the same abilities since CLI command uses the TTS API.

The current example uses the TTS API.

Let's take a look on the synthezis parameters and describe them all.

• The options parameter defines the general configuration of synthezis. Properties:
  • providerName - name of provider name. Available values are: "default" | "custom" | "dasha" | "voice-cloning" | "dasha-emotional"
  • account - the account object of your user. By default the current logged user is used
• The voice parameter is the object that defines the configuration of speech. Properties:
  • lang - using language. For now, only default "en-US"
  • speaker - using speaker. For now, available speakers are: Kate and Linda. The default is Kate.
  • emotion - emotion of synthesed speech. The default is "neutral". To provide custom emotion you have to specify a sample sentence that describes desired emotion in the following format: "from text: your sample setnence text".

Note: to use emotion synthezis you have to set providerName option to "dasha-emotional"

• speed - speed of speach. The default value is 1.
• variation - numeric parameter that affects the pronunciation. The default value is 0. You can play with it to find desirable value.

See the script index.js. It generates several examples of synthesized speech that demonstrates all the parameters. Run it and open output.wav file to hear the difference between them!

Also, see the function synth in the beginning of the script index.js. This function wraps the dasha.tts.synthesize to provide default values of speech synthezis. This values can be overrided by arguments provided from outside. The specified parameters are provided to dasha.tts.synthesize and the resulting synthesized speech (which is array of bytes Uint8Array) is returned. The resulting bytes can be written to a file (like in our case) or can be handled any way you want.

Installation

1. Setup environment (see doc)
2. Run npm i in the current folder.

Running the demo

Run npm start to execute script index.js.