TTS.ai Python SDK

Official Python SDK for the TTS.ai text-to-speech API. Generate high-quality speech from text using 20+ AI models, clone voices, transcribe audio, and more.

Installation

pip install ttsai

For async support:

pip install ttsai[async]

Quick Start

from tts_ai import TTSClient

client = TTSClient(api_key="sk-tts-...")

# Generate speech (returns audio bytes)
audio = client.generate("Hello world!", model="kokoro", voice="af_bella")
with open("output.wav", "wb") as f:
    f.write(audio)

Or set the TTS_API_KEY environment variable:

export TTS_API_KEY=sk-tts-...

client = TTSClient()  # Uses TTS_API_KEY from environment

API Key

Get your API key at tts.ai/account after creating an account.

Usage

Text-to-Speech

The simplest way to generate speech. Uses the OpenAI-compatible /v1/audio/speech endpoint:

from tts_ai import TTSClient

client = TTSClient(api_key="sk-tts-...")

# Basic generation
audio = client.generate("Hello world!")
with open("output.wav", "wb") as f:
    f.write(audio)

# With options
audio = client.generate(
    "Welcome to TTS.ai!",
    model="chatterbox",       # Any supported model
    voice="af_bella",
    output_format="mp3",
    speed=1.2,
)

OpenAI Drop-In Replacement

TTS.ai is compatible with the OpenAI TTS API format:

audio = client.generate(
    "Hello from TTS.ai!",
    model="tts-1",       # Maps to kokoro
    voice="alloy",       # Maps to af_bella
    output_format="mp3",
)

Async TTS (Non-Blocking)

For long-running jobs or when you need the job UUID:

# Start generation (returns immediately)
result = client.generate_async("Long text here...", model="tortoise")
print(f"Job UUID: {result.uuid}")

# Poll for result
audio = client.poll_result(result.uuid, timeout=300)
with open("output.wav", "wb") as f:
    f.write(audio)

Speech-to-Text

# Transcribe from file path
result = client.transcribe("recording.wav")
print(result.text)
print(result.language)

# Transcribe from bytes
with open("recording.wav", "rb") as f:
    audio_bytes = f.read()
result = client.transcribe(audio_bytes, model="faster-whisper")
print(result.text)

Voice Cloning

Clone a voice from a reference audio file (10-30 seconds of clear speech):

result = client.clone_voice(
    name="My Voice",
    file="reference.wav",
    model="chatterbox",    # or cosyvoice2, openvoice, spark, etc.
    text="Hello in my cloned voice!",
)
print(f"Clone job: {result.uuid}")

Supported cloning models: chatterbox, cosyvoice2, glm-tts, gpt-sovits, indextts2, openvoice, spark, tortoise, qwen3-tts.

List Voices and Models

# List all voices
voices = client.list_voices()
for v in voices:
    print(f"{v.voice_id}: {v.name} ({v.language}, {v.gender})")

# Filter by model
kokoro_voices = client.list_voices(model="kokoro")

# List available models
models = client.list_models()
for m in models:
    print(f"{m.name} ({m.tier}): {m.credits_per_1k} credits per 1k chars")

Batch Generation

Process up to 50 texts in a single request:

items = [
    {"text": "First sentence.", "model": "kokoro", "voice": "af_bella"},
    {"text": "Second sentence.", "model": "kokoro", "voice": "af_heart"},
    {"text": "Third sentence.", "model": "chatterbox", "voice": "af_bella"},
]

# Start batch
result = client.batch_generate(items)
print(f"Batch {result.batch_id}: {result.total} items")

# Check progress
status = client.batch_result(result.batch_id)
print(f"Completed: {status.completed}/{status.total}")

# Or generate and wait for completion
result = client.batch_generate_and_wait(items, timeout=300)

With webhook notifications:

result = client.batch_generate(
    items,
    webhook_url="https://yoursite.com/webhook/tts-complete",
)

Context Manager

Use as a context manager to automatically close the HTTP session:

with TTSClient(api_key="sk-tts-...") as client:
    audio = client.generate("Hello!")

Async Client

For asyncio-based applications:

import asyncio
from tts_ai import AsyncTTSClient

async def main():
    async with AsyncTTSClient(api_key="sk-tts-...") as client:
        # All methods are async
        audio = await client.generate("Hello world!")
        with open("output.wav", "wb") as f:
            f.write(audio)

        # Parallel generation
        tasks = [
            client.generate("First sentence.", voice="af_bella"),
            client.generate("Second sentence.", voice="af_heart"),
        ]
        results = await asyncio.gather(*tasks)

asyncio.run(main())

Error Handling

from tts_ai import TTSClient
from tts_ai.exceptions import (
    AuthenticationError,
    InsufficientCreditsError,
    RateLimitError,
    ModelNotFoundError,
    TimeoutError,
    TTSError,
)

client = TTSClient(api_key="sk-tts-...")

try:
    audio = client.generate("Hello!", model="kokoro")
except AuthenticationError:
    print("Invalid API key")
except InsufficientCreditsError as e:
    print(f"Not enough credits (have {e.credits_remaining}, need {e.credits_needed})")
except RateLimitError:
    print("Rate limit exceeded, try again later")
except ModelNotFoundError:
    print("Model not found")
except TimeoutError:
    print("Request timed out")
except TTSError as e:
    print(f"API error ({e.status_code}): {e.message}")

Available Models

Model	Tier	Cloning
kokoro	Free	No
piper	Free	No
vits	Free	No
melotts	Free	No
chatterbox	Premium	Yes
cosyvoice2	Standard	Yes
bark	Standard	No
dia	Standard	No
glm-tts	Standard	Yes
gpt-sovits	Standard	Yes
indextts2	Standard	Yes
openvoice	Premium	Yes
orpheus	Standard	No
parler	Standard	No
qwen3-tts	Standard	Yes
sesame-csm	Premium	No
spark	Standard	Yes
styletts2	Premium	No
tortoise	Premium	Yes

Use client.list_models() for the latest list with credit costs.

Configuration

Parameter	Environment Variable	Default
api_key	TTS_API_KEY	(required)
base_url	-	https://tts.ai
gpu_url	-	https://api.tts.ai
timeout	-	120 seconds
max_retries	-	3

License

MIT