ElevenLabsLib - Unofficial API wrapper#

Note

For additional logging with this library, import logging and set the logging level to DEBUG.

Migrating from 0.12 to 0.13+#

ElevenLabsVoice now caches most of the information on a voice, such as settings, name, description, once the instance is created.

This means that methods such as voice.get_settings or voice.get_name, which hit the API every time, are deprecated.

The new approach is to use the provided properties, and use voice.update_data to update all voice data if you want to ensure you have the latest data.

In practical terms, this:

voice = user.get_voice_by_id("voiceID")
print(voice.get_settings())

Becomes:

voice = user.get_voice_by_id("voiceID")
print(voice.settings)

And if you want to make sure you have the latest settings:

voice = user.get_voice_by_id("voiceID")
voice.update_data()
print(voice.settings)

In almost all instances, once you’ve created a voice object all the associated information is immediately cached as part of its creation.

This means that calling a property will never hit the API - with one exception detailed below.

Warning

The /voices API endpoint is currently bugged, and does not return settings information.

This means that any voice object created by calling user.get_all_voices (or user.get_available_voices) will be missing settings information.

In this case, the first call to voice.settings will actually hit the API endpoint to get the required information.

Migrating from 0.9 to 0.10+#

In the 0.10 release, many functions that deal with audio generation/playback have been deprecated, and new _v2 versions have been introduced.

Instead of specifying each option as a separate argument, they are now grouped together into two dataclasses, PlaybackOptions and GenerationOptions.

For now the old versions are merely deprecated but they will be removed in the future. Additionally, it is not possible to leverage the new settings of the V2 english model with the deprecated functions.

Here’s an example of how to migrate, using generate_and_stream_audio as an example.

The old function signature looked like this:

generate_stream_audio(prompt, portaudioDeviceID=None, stability=None,
                      similarity_boost=None, streamInBackground=False,
                      onPlaybackStart=lambda: None, onPlaybackEnd=lambda: None,
                      model="eleven_monolingual_v1", latencyOptimizationLevel=0)

Now, the new function signature looks like this:

generate_stream_audio_v2(prompt, playbackOptions, generationOptions)

Note

Similar changes apply to many other functions.

To migrate your existing code, follow these steps:

  1. Import the necessary dataclasses:

    from elevenlabslib import PlaybackOptions, GenerationOptions

  2. Create instances of PlaybackOptions and GenerationOptions with the appropriate settings:

    playbackOptions = PlaybackOptions(runInBackground=True, portaudioDeviceID=1)
generationOptions = GenerationOptions(model="new_model_id")

  3. Call the new functions with your text prompt and the options instances:

    result = voice.generate_stream_audio_v2("Hello, world!",
                                             playbackOptions,
                                             generationOptions)

Please refer to the PlaybackOptions and GenerationOptions sections for more information about the available settings in these dataclasses.