NVIDIA Riva

​

Overview

RivaTTSService converts text to speech using NVIDIA’s Riva. It provides high-quality text-to-speech synthesis with configurable voice options, including multilingual voices.

​

Installation

To use RivaTTSService, install the required dependencies:

pip install "pipecat-ai[riva]"

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

​

Configuration

​

Constructor Parameters

​

InputParams

​

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.

​

Models

​

Language Support

Riva model magpie-tts-multilingual (default) supports English, Spanish, and French:

​

Usage Examples

​

TTS Language and Voice Configuration

from pipecat.services.riva.tts import RivaTTSService
from pipecat.transcriptions.language import Language

# Configure service
tts = RivaTTSService(
    api_key="your-nvidia-api-key",
    voice_id="Magpie-Multilingual.FR-FR.Louise",
    params=RivaTTSService.InputParams(
        language=Language.FR_FR,
        quality=20
    )
)

# Use in pipeline
pipeline = Pipeline([
    ...,
    llm,
    tts,
    transport.output(),
])

​

Model, function ID, and Voice configuration

# Configure TTS with specific language
tts = RivaTTSService(
    api_key="your-nvidia-api-key",
    voice_id="English-US.Female-1",
    model_function_map={
      "function_id": "0149dedb-2be8-4195-b9a0-e57e0e14f972",
      "model_name": "fastpitch-hifigan-tts",
    }
)

# Use in pipeline
pipeline = Pipeline([
    ...,
    llm,
    tts,
    transport.output(),
])

​

Frame Flow

​

Metrics Support

The service supports metrics collection:

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

​

Audio Processing

• Processes audio through the Riva API
• Generates mono audio output
• Handles asynchronous audio streaming
• Configurable sampling rate

​

Notes

• Uses NVIDIA’s Riva AI Services platform
• Streams audio in chunks
• Requires valid NVIDIA API key
• Thread-safe processing with asyncio