ElevenLabs

The audio output format for the TTS audio. This is a string that specifies the format and sample rate of the audio. The format is pcm_<sample_rate>.

The sample rate must be one of these values: pcm_16000, pcm_22050, pcm_24000, pcm_44100.

{
  "service_options": {
    "elevenlabs": {
      "output_format": "pcm_24000"
    }
  }
}

Select from the available ElevenLabs models. For the lowest latency, we recommend using the Turbo V2.5 model ("eleven_turbo_v2_5").

{
  "name": "model",
  "value": "eleven_turbo_v2_5"
}

The voice you want your TTS service to use. Select any voice from the available ElevenLabs voices.

{
  "name": "voice",
  "value": "Xb7hH8MSUJpSbSDYk0k2"
}

The language you want your TTS service to use. Learn more.

{
  "name": "language",
  "value": "es"
}

You can turn on latency optimizations at some cost of quality. The best possible final latency varies by model.

Values range from 0 (default) to 4 (most aggressive).

{
  "name": "optimize_streaming_latency",
  "value": "3"
}

Defines the stability for voice settings. Range is from 0 to 1.

{
  "name": "stability",
  "value": 0.5
}

Defines the similarity boost for voice settings. Range is from 0 to 1.

{
  "name": "similarity_boost",
  "value": 0.5
}

Defines the style for voice settings. Range is from 0 to 1.

{
  "name": "style",
  "value": 0.5
}

Defines the use speaker boost for voice settings.

{
  "name": "use_speaker_boost",
  "value": "true"
}

Control whether the TTS service filters out markdown, code blocks, or tables from its output.

Basic markdown filtering is enabled by default. Enable code and table filtering as needed. Filtering code and tables can help the TTS avoid mistakes.

{
  "name": "text_filter",
  "value": {
    "enable_text_filter": true,
    "filter_code": true,
    "filter_tables": true
  }
}