Directory/Audio Generation/AI Music Generation/AudioCraft

AudioCraft

AudioCraft is Meta AI's research codebase for generative audio, bundling three models behind one PyTorch library. MusicGen turns text (and optional melody) prompts into music; AudioGen generates environmental sound effects from text; EnCodec is the neural audio codec that converts raw waveforms into parallel streams of discrete tokens and back. Architecturally it's a single autoregressive language model operating over EnCodec tokens with a token-interleaving pattern that captures long-range structure, conditioned by a pretrained text encoder. That means one model, not a pipeline of separately trained stages — easier to fine-tune, easier to reason about. Built for ML researchers, audio engineers, and game/media developers who can run Python and own a GPU. It ships as code, not a SaaS product: you clone the repo, download the model cards, and generate locally. Expect non-commercial licensing on the weights and the usual research-grade rough edges.

Audio Generation

Visit website

Added on

Nov 4, 2026

MusicGenAudioGenEnCodecText-to-musicText-to-soundAudio compressionOpen sourceMeta AI research

Product information

Everything worth knowing about AudioCraft

The complete picture — from everyday features to the technical detail builders and IT folks dig for.

Core features

What it actually does

Text-to-Music Generation

MusicGen turns text prompts into diverse, long music samples — no DAW, no sample-hunting, just describe what you want.

Text-to-Sound Generation

AudioGen generates environmental sound effects from text descriptions, aimed at foley and sound design work.

Neural Audio Compression

EnCodec compresses raw waveforms into compact discrete representations and reconstructs them, handling the encode/decode round trip for all AudioCraft models.

Open-Source Codebase + Model Cards

Everything ships as code on GitHub with published model cards, so you can inspect, run, and reproduce the research yourself.

Technical capabilities

Single Autoregressive LM Architecture

Both MusicGen and AudioGen are one autoregressive language model operating over streams of compressed discrete audio tokens, rather than separate multi-stage pipelines.

Parallel Token Stream Interleaving

Uses an explicit token interleaving pattern across EnCodec's parallel token streams, letting one model capture long-term audio dependencies while generating high-quality output.

Pluggable Conditioning Models

Generation is steered by conditioning modules such as a pretrained text encoder for text-to-audio, keeping the core LM decoupled from the control signal.

Must watch videos

How To Install Audiocraft Locally - Meta's FREE And Open AI Music Gen

5:14

Tutorial5:14

How To Install Audiocraft Locally - Meta's FREE And Open AI Music Gen

In this video, we'll learn how to install Audiocraft, the new open-source and free AI music and sound effect generator from Meta AI. I go over both Musicgen and Audiogen. Enjoy! Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.ai/ My Links 🔗 👉🏻 Subscribe: https://www.youtube.com/@matthew_berman 👉🏻 Twitter: https://twitter.com/matthewberman 👉🏻 Discord: https://discord.gg/xxysSXBxFW 👉🏻 Patreon: https://patreon.com/MatthewBerman Media/Sponsorship Inquiries 📈 https://bit.ly/44TC45V Chapters: 0:00 - Intro 0:11 - Musicgen Installation 3:13 - Audiogen Installation Links: Install Instructions: https://gist.github.com/mberman84/afd800f8d4a8764a22571c1a82187bad Audiocraft Github - https://github.com/facebookresearch/audiocraft Audiogen PR - https://github.com/facebookresearch/audiocraft/pull/185 Audiocraft Announcement - https://about.fb.com/news/2023/08/audiocraft-generative-ai-for-music-and-audio/

Easy AI Sound Effects - AudioGen Tutorial - Audiocraft Plus

8:04

Tutorial8:04

Easy AI Sound Effects - AudioGen Tutorial - Audiocraft Plus

#ai #aitutorial #audioengineer #soundeffects #audiogen #audiocraft Learn how to create AI sound effects locally on your PC by using Audiogen (Audiocraft Plus). Besides single sound effects you can also create a sequence of sound effects and edit existing sound effects. At the end of the video, I’ll also briefly show you some basics in animating images. Links Download Pinokio (One-Click-Installer for AI tools): https://pinokio.computer Photomator for Mac and iOS https://www.pixelmator.com/photomator/ Motionleap for iOS and Android https://www.lightricks.com/products In case you wonder what inspired the airplane scene: https://www.reddit.com/r/UFOs/comments/15p14tp/megathread_mh370_relevant_posts_regarding_mh370/ Audiogen Prompts Airplane scene: “Airplane landing” “Thunder” “Airplane” City Car scene: “Heavy rain“ “Starting a cheap car“ “Sound of a lamborghini“ Hangar scene: “UFO hovering humm” Fireplace: “Fire crackling” Cow scene: “Birds chirping in the background” “Cow eating grass” “UFO plasma beam” “Cow in distress” Cassette Player: “Turning on a cassette player” Chapters 00:00 Intro 00:30 Overview 00:49 Install Audiogen (Audiocraft Plus) 01:19 Generate AI Sound Effects 02:36 Comparison of different Seeds 03:09 Comparison of different Prompts with the same Seed 03:25 Generate a Sequence of AI Sound Effects 04:43 Edit existing Sound Files 05:55 Customization and Settings 06:20 Read out the Prompt of generated Sound Files 06:30 Animate Images in Video Editing Software 07:52 Outro

MusicGen Tutorial – Remix Any Song with this free AI Music Maker – Audiocraft Plus

9:52

Tutorial9:52

MusicGen Tutorial – Remix Any Song with this free AI Music Maker – Audiocraft Plus

#aimusic #aitutorial #ai #musicgen #aisongs #aisong Learn how to generate AI music locally on your PC by using MusicGen (Audiocraft Plus). Besides normal music, you can also create a sequence of music or even create AI remixes of any existing songs. Links: Download Pinokio (which lets you install Audiocraft+ with one click): https://pinokio.co Generate AI Sound Effects: https://www.youtube.com/watch?v=Gue4vfouSFU Collection of sample prompts for inspiration: https://ai.honu.io/papers/musicgen/ Dig Deep – RW Smith: https://www.youtube.com/watch?v=MF_zoKQysfI Chapters: 00:00 Intro 00:28 Overview 00:41 Install MusicGen (Audiocraft Plus) 01:01 Generate Music with MusicGen 01:36 Improve the Music Generation 02:14 MusicGen Settings explained 03:59 Generate a Music Sequence 06:11 Location of generated Files 06:25 Remix Songs in MusicGen 06:59 Remix Songs with the Sample Method 07:52 Remix Songs with the Melody Method 08:44 Read Settings from generated File 09:03 Examples 09:32 Outro

Who uses it

Real use cases, no hype

Text-to-Music Generation

Type a prompt like 'lo-fi jazz with brushed drums' and get a full track from MusicGen.

Sound Effect Design

Generate environmental sounds — rain, footsteps, engines — with AudioGen instead of licensing sample packs.

Melody-Conditioned Composition

Feed a hummed or whistled melody and let MusicGen build a full arrangement around it.

Neural Audio Compression

Use EnCodec standalone to squeeze audio down to 1.5–24 kbps for streaming or storage.

Model Fine-Tuning

Train MusicGen on your own catalog with the bundled Dora pipeline for a custom style model.

Game & App Prototyping

Ship placeholder audio assets fast without waiting on a sound designer.

Audio Research Baseline

Fork the codebase as a reproducible baseline for your own generative audio experiments.

What's great

Free, open-source audio models from Meta AI
MusicGen + AudioGen + EnCodec in one repo
Runs fully local — no API keys, no per-minute fees
Melody conditioning: hum a tune, get an arrangement

Technical strengths

Elegant design: one autoregressive LM over discrete EnCodec tokens
Token interleaving models long audio without ballooning sequence length
Full training and Dora fine-tuning code, not just inference

Where it falls short

Research repo, not a product — no hosted UI
Hungry for VRAM; large model wants a serious GPU
Docs thin on fine-tuning, dataset prep, and failure modes
Model weights are CC-BY-NC — commercial use blocked

Technical limitations

Autoregressive sampling is slow; no real-time streaming
EnCodec introduces high-frequency artifacts and muffled tails
Brittle installs — pinned old PyTorch/torchaudio versions

For technical folks

The deep-dive specs

Architecture

Single autoregressive Language Model (LM) over compressed discrete audio tokens

Audio Representation

Parallel streams of discrete tokens produced by EnCodec

Codec

EnCodec neural audio codec (encoder/decoder for waveform ↔ token mapping)

Sequence Modeling

Token interleaving pattern across parallel streams for long-term dependency capture

Conditioning

Pluggable conditioning models, e.g. pretrained text encoder for text-to-audio

Tasks Covered

Music generation, environmental sound generation, audio compression

Distribution

Open-source code on GitHub with published model cards

FAQ

Includes technical Q&A

Traffic Insights

Monthly Visits

—

November 2026

Growth Rate

vs last month

Dominance

in Audio Generation

Top Country

—

Top Countries Breakdown

—

Growth Trend

Stable

Traffic has remained stable.

Pricing overview

Unknown Pricing

AudioCraft Pricing

AudioCraft uses a unknown pricing structure. Real-time rates, custom enterprise quotas, and plan details are managed directly on their official platform.

Check Pricing on AudioCraft

Reviews