← All nodes
Transforms a_tts_stream

Text-to-Speech Stream

Streams text-to-speech audio chunks via a TTS provider (Deepgram).

What does it do?

Converts text to speech in real time, sending audio piece by piece.

How does it work?

Generates audio as the AI is composing the reply, streaming it without delays.

Why use it?

Exclusive for phone bots. Reduces latency (wait time on the call) so it feels like talking to a real human.

Inputs

nametype
text string

Outputs

nametype
url string
provider string
model string
format string
sample_rate number
bytes number
chunks number
duration_ms number
chars number

Branches

Fixed branches

  • out — Output

Configuration

model string optional
voice string optional
text_field string required default: "text"
format "mp3" | "pcm" | "mulaw" required default: "mp3"
sample_rate integer required default: 24000
persist_to_r2 boolean required default: false
memory_key string optional

Default configuration

{
  "text_field": "text",
  "format": "mp3",
  "sample_rate": 24000,
  "persist_to_r2": false
}

Emitted events

  • node.token

    Streaming TTS audio tokens.

  • tts.done

    TTS synthesis completed.

  • tts.error

    TTS synthesis error.