Up to 100x Cheaper, High Quality, Realtime Text-to-Speech
Sign Up for an API Key »
Read the docs
·
Report Bug
- Easy installation with pip.
- Streaming audio bytes and non-streaming both supported.
- Async/await and standard sync code both supported.
- Full IDE support with autocomplete, type-hinting, and in-code documentation.
- Installation
- Basic Example
- Environment Variables
- Streaming Example
- Async Support
- Voice Options
- Output Formats
- Voice API
Install the python package as a dependency of your application.
$ pip install mythicinfinityfrom mythicinfinity import MythicInfinityClient
def main():
# Instantiate the client with your api key
client = MythicInfinityClient(api_key="YOUR_API_KEY")
# Call the TTS API. By default, stream is False.
audio_bytes = client.tts.generate(text="Hello world.")
with open('my_audio.wav', 'wb') as f:
f.write(audio_bytes)
if __name__ == "__main__":
main()This sample calls the Text-To-Speech service and saves the resultant audio bytes to a file.
| Name | Description |
|---|---|
MYTHICINFINITY_API_KEY |
Sets the api key. There is no need to pass api_key="YOUR_API_KEY" to the client constructor when using the environment variable. |
from mythicinfinity import MythicInfinityClient
def main():
# Instantiate the client with your api key
client = MythicInfinityClient(api_key="YOUR_API_KEY")
# Call the TTS API with stream=True.
# This will stream the audio bytes in real-time,
# as they become available from the AI model.
audio_bytes_generator = client.tts.generate(text="Hello world.", stream=True)
with open('my_audio.wav', 'wb') as f:
for audio_bytes in audio_bytes_generator:
f.write(audio_bytes)
if __name__ == "__main__":
main()- Code relying on
async / awaitpatterns is fully supported.
import asyncio
from mythicinfinity import AsyncMythicInfinityClient
async def main():
# Instantiate the client with your api key
client = AsyncMythicInfinityClient(api_key="YOUR_API_KEY")
# Call the TTS API with stream=False.
audio_bytes = await client.tts.generate(text="Hello world.")
with open('my_async_audio_1.wav', 'wb') as f:
f.write(audio_bytes)
# Call the TTS API with stream=True.
# This will stream the audio bytes in real-time,
# as they become available from the AI model.
audio_bytes_generator = await client.tts.generate(text="Hello world.", stream=True)
with open('my_async_audio_2.wav', 'wb') as f:
async for audio_bytes in audio_bytes_generator:
f.write(audio_bytes)
if __name__ == "__main__":
asyncio.run(main())This sample first calls the client.tts.generate method without streaming using await and then does the same with
streaming enabled.
The voice_id parameter controls which voice is used for the generated audio.
You can use the Voice API to list all available voice ids and get their related metadata.
The model supports various options which can be set to adjust the way audio is generated with the voice in various ways.
Each voice may specify its own default values for these parameters.
Voice Option Parameters:
consistency- The recommended value is 2.0. Higher values tend to follow text more reliably, and pronounce words with more accuracy, but alters the way speech is spoken and may increase the speaking rate.
from mythicinfinity import MythicInfinityClient, VoiceOptions
def main():
# Instantiate the client with your api key
client = MythicInfinityClient(api_key="YOUR_API_KEY")
# Call the API setting the consistency VoiceOption
audio_bytes_generator = client.tts.generate(
text="Hello world.", stream=True,
voice_id="",
voice_options=VoiceOptions(
consistency=2.4,
)
)
with open('my_audio.wav', 'wb') as f:
for audio_bytes in audio_bytes_generator:
f.write(audio_bytes)
if __name__ == "__main__":
main()By default, outputs are in the wav format.
The currently supported output formats are:
wav- 24khz 16 bit WAVpcm- 24khz 16 bit raw PCM datamp3- 24khz 192kbps MP3webm_opus- 24khz 128kbps OPUS inside a WEBM container (saved as a.webmfile)
For streaming playback (that starts before the whole audio is done generating), usually mp3 or webm_opus are used.
For the highest output quality, we recommend wav.
To specify the output format, simply set the format parameter in the .generate() call.
Valid values are wav, pcm, mp3 and webm_opus.
from mythicinfinity import MythicInfinityClient
def main():
# Instantiate the client with your api key
client = MythicInfinityClient(api_key="YOUR_API_KEY")
# Call the API setting format=mp3
audio_bytes_generator = client.tts.generate(text="Hello world.", stream=True, format="mp3")
# Save the audio as a .mp3 file
with open('my_audio.mp3', 'wb') as f:
for audio_bytes in audio_bytes_generator:
f.write(audio_bytes)
if __name__ == "__main__":
main()class Voice:
name: str
voice_id: str
model_ids: typing.List[str]
"""
Model IDs that this voice is compatible with.
"""
preview_urls_by_model_id: typing.Dict[str, str]
"""
Preview urls for this voice per model id.
"""The voice objects returned by this api will have this structure.
Sync
all_voices = client.tts.voices.list()Async
all_voices = await async_client.tts.voices.list()Sync
voice = client.tts.voices.get("kiera")Async
voice = await async_client.tts.voices.get("kiera")