Cartesia Voice Provider

Cartesia provides ultra-low-latency text-to-speech synthesis through the Sonic voice engine. The Beluga AI provider uses Cartesia’s HTTP API with the httpclient infrastructure for built-in retry support, producing raw PCM audio output suitable for real-time voice pipelines.

Choose Cartesia when latency is critical — for example, in conversational voice agents where every millisecond of TTS delay affects the user experience. Cartesia’s Sonic engine is optimized for speed-first synthesis with direct PCM output, avoiding the overhead of compressed audio decoding. For the highest voice quality with more voice variety, consider ElevenLabs.

Installation

import _ "github.com/lookatitude/beluga-ai/voice/tts/providers/cartesia"

The blank import registers the "cartesia" provider with the TTS registry.

Configuration

Field	Type	Default	Description
Voice	string	—	Cartesia voice UUID
Model	string	"sonic-2"	Cartesia model (sonic-2, sonic-english)
SampleRate	int	24000	Output sample rate in Hz
Extra	—	—	See below

Extra Fields

Key	Type	Required	Description
api_key	string	Yes	Cartesia API key
base_url	string	No	Override base URL

Basic Usage

package main

import (
    "context"
    "log"
    "os"

    "github.com/lookatitude/beluga-ai/voice/tts"
    _ "github.com/lookatitude/beluga-ai/voice/tts/providers/cartesia"
)

func main() {
    ctx := context.Background()

    engine, err := tts.New("cartesia", tts.Config{
        Voice: "a0e99841-438c-4a64-b679-ae501e7d6091",
        Extra: map[string]any{"api_key": os.Getenv("CARTESIA_API_KEY")},
    })
    if err != nil {
        log.Fatal(err)
    }

    audio, err := engine.Synthesize(ctx, "Hello, welcome to Beluga AI.")
    if err != nil {
        log.Fatal(err)
    }

    if err := os.WriteFile("output.pcm", audio, 0644); err != nil {
        log.Fatal(err)
    }
}

Direct Construction

import "github.com/lookatitude/beluga-ai/voice/tts/providers/cartesia"

engine, err := cartesia.New(tts.Config{
    Voice:      "a0e99841-438c-4a64-b679-ae501e7d6091",
    Model:      "sonic-2",
    SampleRate: 24000,
    Extra:      map[string]any{"api_key": os.Getenv("CARTESIA_API_KEY")},
})

Streaming

The streaming interface synthesizes each text chunk from the input stream independently:

for chunk, err := range engine.SynthesizeStream(ctx, textStream) {
    if err != nil {
        log.Printf("error: %v", err)
        break
    }
    transport.Send(chunk)
}

FrameProcessor Integration

processor := tts.AsFrameProcessor(engine, 24000, tts.WithVoice("a0e99841-438c-4a64-b679-ae501e7d6091"))
pipeline := voice.Chain(sttProcessor, llmProcessor, processor)

Advanced Features

Output Format

Cartesia outputs raw PCM audio (16-bit little-endian, pcm_s16le) by default. The output format is configured in the request body and matches the sample rate specified in the config:

engine, err := tts.New("cartesia", tts.Config{
    Voice:      "a0e99841-438c-4a64-b679-ae501e7d6091",
    SampleRate: 44100, // Override default 24000 Hz
    Extra:      map[string]any{"api_key": os.Getenv("CARTESIA_API_KEY")},
})

Built-in Retry

The Cartesia provider uses Beluga’s httpclient infrastructure, which provides automatic retry with exponential backoff (up to 2 retries by default) for transient failures.

Per-Request Options

audio, err := engine.Synthesize(ctx, "Hello!",
    tts.WithVoice("different-voice-uuid"),
    tts.WithSampleRate(16000),
)

Custom Endpoint

engine, err := tts.New("cartesia", tts.Config{
    Voice: "a0e99841-438c-4a64-b679-ae501e7d6091",
    Extra: map[string]any{
        "api_key":  os.Getenv("CARTESIA_API_KEY"),
        "base_url": "https://cartesia.internal.corp",
    },
})