Proof of Use. An independent field guide · No. 01
Back to the collection
Music & Audio

Chatterbox

Generate expressive speech from text with Resemble AI’s local text-to-speech family, including multilingual and reference-voice options.

text to speechnarrationmultilingualvoice
Source-reviewedNot hands-on testedSources checked

Start with an idea

What you can make

01

Prototype narration or spoken interfaces from your own scripts

02

Create multilingual voice prototypes using a voice you have permission to use

Creator demonstration

See the documented workflow

Creator-curated expressive, voice-agent and multilingual samples; playback and model output were not independently tested.

Hear Resemble AI’s speech examples

Before you begin

What it takes to run

Python + model download

  • Upstream reports testing with Python 3.11 on Debian 11; current package metadata requires Python 3.10 or later
  • Install the PyTorch/audio dependency stack and download the selected model checkpoints
  • CUDA examples and a Mac MPS/CPU example are provided; choose hardware for the exact variant and measure performance
  • Reference-voice synthesis requires an audio clip you have rights and permission to use
Runs through / on

Linux (Debian tested upstream) · macOS example (MPS / CPU) · Python · Local Gradio interface

The real tradeoffs

Cost & licensing

Running costs

No license fee for self-hosted use; local or rented inference compute is your cost.

Resemble’s hosted TTS service is a separate paid offering. No hosted rate, latency guarantee or universal minimum VRAM has been independently verified here.

Code license

MIT

Read the code license

Models & other terms

MIT, as declared in the official base and Turbo model cards

Read the model terms

The actual GitHub LICENSE was inspected. The base weights repository declares MIT in its model card but contains no separate LICENSE file; verify any separately downloaded variant. Voice permission and other rights remain separate.

Know the limitations

  • Language coverage and controls differ between English, multilingual and smaller variants
  • Reference audio can affect accent and pacing; listen to output before publishing
  • Generated audio includes the project’s PerTh watermark
  • Creator performance claims and demo samples are not independent measurements

Follow the evidence

Sources & verification

Official sources were reviewed on October 14, 2026. We did not install, run or benchmark this project. This is a source-check date, not a release date.

Links and project terms can change. These sources document the claims above; they are not a safety certification. Read our full method.

Keep exploring

A few other possibilities

All tools

05Music & Audio

Whisper

Transcribe speech in multiple languages or translate spoken audio into English with OpenAI’s locally runnable speech-recognition models.

Try it for

Create a reviewable transcript of your own interviews or podcasts

Read entry Official examples
Setup · Moderate
Python + FFmpeg
Cost to consider
Local compute

06Music & Audio

AudioCraft / MusicGen

Explore text- and melody-conditioned music generation with Meta’s AudioCraft research toolkit and MusicGen models.

Try it for

Sketch instrumental musical ideas for noncommercial experimentation

Read entry Official examples
Setup · Advanced
Python + GPU setup
Cost to consider
GPU compute; noncommercial weights

08Music & Audio

Demucs

Separate a mixed music recording into vocals, drums, bass and other accompaniment with a locally runnable source-separation model.

Try it for

Isolate parts from your own recordings for practice and analysis

Read entry Official examples
Setup · Moderate
Python + audio dependencies
Cost to consider
Local compute; check weights terms