OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Audio & Voice AI

Audiolm Pytorch

Audiolm Pytorch is a Python and PyTorch implementation of AudioLM that extends the model with T5 conditioning and classifier-free guidance for text-to-audio and text-to-speech experiments.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Audiolm Pytorch — OpenViST project profile in Audio & Voice AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want a code-level starting point for AudioLM experiments or need the repository’s documented SoundStream and EnCodec audio-codec workflows.

Who it is for

  • Developers building or adapting audio-generation workflows in Python.
  • Researchers experimenting with AudioLM, text-conditioned speech and audio codecs.

Documented uses

  • Experiment with generating audio from text conditioning.
  • Explore text-to-speech model workflows using T5 conditioning.
  • Train SoundStream on an audio corpus and use the trained model to tokenize or reconstruct audio.
  • Use the documented EnCodec wrapper with pretrained 24 kHz EnCodec.
Before you choose

Requirements & limits

The repository is public and MIT-licensed. The README gives the package installation command `pip install audiolm-pytorch`.

  • The repository adds T5 conditioning with classifier-free guidance, extending its AudioLM implementation to text-to-audio and text-to-speech experiments.
  • The README offers two codec routes: its SoundStream implementation, which requires training on a large audio corpus, or an EnCodec wrapper for pretrained 24 kHz EnCodec.
  • The SoundStream example is a developer workflow using local audio files and CUDA, not a hosted audio-generation service.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up Audiolm Pytorch

Install the Python package so its documented audio components can be imported.

Before you start

  • Python environment with pip

Steps

  1. Run the command below.

    Documented command
    pip install audiolm-pytorch

  2. Import the components needed for your workflow, such as EncodecWrapper, SoundStream or SoundStreamTrainer.

Expected outcome

The package is installed and the README’s audio components are available to import in Python.

Guide limitations

  • Installation alone does not train a model or generate audio.
Documentation 1 ↗
Source-based practical guide

Prepare a SoundStream training workflow

Configure the documented SoundStream trainer to learn an audio codec from local audio files.

Before you start

  • The installed audiolm-pytorch package
  • A large corpus of audio files in a folder
  • A CUDA device for the README’s shown trainer configuration

Steps

  1. Create a SoundStream model with the desired codec parameters.

  2. Pass the model and audio folder to SoundStreamTrainer, setting training parameters such as batch size, maximum audio length and number of steps.

  3. Move the trainer to CUDA and call trainer.train().

Expected outcome

The documented workflow trains a SoundStream model that can then be used as an audio tokenizer or for audio reconstruction.

Guide limitations

  • The README says SoundStream needs training on a large audio corpus.
  • The example uses one million training steps; it does not establish a training duration or resource guarantee.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
lucidrains/audiolm-pytorch · Official homepage title · retained in its source language
How we review sourcesSuggest a correction