Deepgram

​

Overview

DeepgramTTSService converts text to speech using Deepgram’s Aura API. It supports various voices and audio configurations.

​

Installation

To use DeepgramTTSService, install the required dependencies:

pip install pipecat-ai[deepgram]

You’ll also need to set up your Deepgram API key as an environment variable: DEEPGRAM_API_KEY

​

Configuration

​

Constructor Parameters

​

Input

The service accepts text input through its TTS pipeline.

​

Output Frames

​

TTSStartedFrame

Signals the start of audio generation.

​

TTSAudioRawFrame

Contains generated audio data:

​

TTSStoppedFrame

Signals the completion of audio generation.

​

Methods

See the TTS base class methods for additional functionality.

​

Language Support

Deepgram TTS supports the following languages and regional variants:

​

Usage Example

from pipecat.services.deepgram import DeepgramTTSService

# Configure service
tts_service = DeepgramTTSService(
    api_key="your-api-key",
    voice="aura-helios-en",
    sample_rate=24000
)

# Use in pipeline
pipeline = Pipeline([
    text_input,         # Produces text for synthesis
    tts_service,        # Converts text to speech
    audio_output        # Handles audio playback
])

​

Frame Flow

​

Metrics Support

The service supports metrics collection:

• Time to First Byte (TTFB)
• TTS usage metrics
• Processing duration

​

Audio Processing

• Streams audio in 8KB chunks
• Supports 16-bit PCM format
• Generates mono audio output
• Handles memory buffering

​

Error Handling

try:
    # Generate speech
    async for frame in service.run_tts(text):
        if isinstance(frame, ErrorFrame):
            handle_error(frame.error)
except Exception as e:
    logger.error(f"TTS error: {e}")

​

Notes

• Requires valid Deepgram API key
• Streams audio in chunks
• Supports various voices
• Provides metrics collection
• Handles memory efficiently
• Thread-safe processing