What can I use LFM3-Audio-2B for?

Give it a .wav file with the instruction to perform speech recognition in English, and you get a written transcript. Give it a sentence of text with the instruction to speak in an American English female voice, and you get spoken audio. Give it a recording and ask what sounds or events it contains, and you get a written answer about the audio.

Can I run LFM3-Audio-2B on my computer?

It runs on your own hardware with a model runtime. The model card mentions Transformers, vLLM, SGLang; check its setup and hardware notes.

What do I need to use LFM3-Audio-2B?

Python with the Transformers library and torch. Other builds exist for llama.cpp on CPUs and GPUs, MLX-VLM on Apple Silicon, and ONNX Runtime. Memory needs are not documented.

LFM3-Audio-2B: Transcribe, speak and chat by voice (2B)

Model details

Original model LiquidAI/LFM3-Audio-2B

Capability unclear

Input Unknown Output Unknown

Follow the model card setup

Ways to use it Run locally

What it does

LFM3-Audio-2B is a compact audio model that takes text or audio as input and replies with text, speech or both. It can transcribe, synthesize and translate speech across 18 languages, and answer questions about the speech, sounds and events in a recording. You choose the task with a short system prompt.

How to use it

Load the weights with the Transformers library from the Hugging Face page, then give the model audio or text plus a system prompt such as the transcription or speech-synthesis instruction. A GGUF version runs with llama.cpp, and a 4-bit MLX version runs on Apple Silicon.

Common use cases

Give it a .wav file with the instruction to perform speech recognition in English, and you get a written transcript.
Give it a sentence of text with the instruction to speak in an American English female voice, and you get spoken audio.
Give it a recording and ask what sounds or events it contains, and you get a written answer about the audio.

Can I use this?

Needs a model runtime on your computer

Requirements It runs on your own computer or edge device; install Transformers, torch, accelerate, librosa and soundfile, then load the model with its custom code enabled.

Caveat Text-only answers need the mixed-output and audio-output suppress options turned on, and speech output works best with the sampling settings the source lists.

Worth knowing Live speech recognition currently supports English only. It is not recommended for agentic coding or complex multi-step reasoning. Licensed under the LFM Open License v1.0.

Why it's here

Not trending this week

Last seen on the Hugging Face trending list on 3 November 2026. It stays on BuildTube so you can still find it.

245 people have liked it on Hugging Face.
It was downloaded 10,465 times in the last 30 days.

Numbers checked on 8 November 2026; not refreshed since.

Behind it

View model page

BuildTube has not run or verified this project. Everything above is written from what the creator published.

Made this?

Quick questions

What is LFM3-Audio-2B?: LFM3-Audio-2B is a compact audio model that takes text or audio as input and replies with text, speech or both.
What can I use LFM3-Audio-2B for?: Give it a .wav file with the instruction to perform speech recognition in English, and you get a written transcript. Give it a sentence of text with the instruction to speak in an American English female voice, and you get spoken audio. Give it a recording and ask what sounds or events it contains, and you get a written answer about the audio.
Can I run LFM3-Audio-2B on my computer?: It runs on your own hardware with a model runtime. The model card mentions Transformers, vLLM, SGLang; check its setup and hardware notes.
What do I need to use LFM3-Audio-2B?: Python with the Transformers library and torch. Other builds exist for llama.cpp on CPUs and GPUs, MLX-VLM on Apple Silicon, and ONNX Runtime. Memory needs are not documented.
What license does LFM3-Audio-2B use?: Licensed under the LFM Open License v1.0.

More from this week

P-H3-Image-Comfy Make and edit 8-megapixel images in ComfyUI
Turn short text requests into 8-megapixel images and edit pictures on your own graphics card.
Added 7 Nov
Idol Turn one picture of an object into a 3D Gaussian splat
Idol turns one photo, render or generated image of an object into a 3D Gaussian splat file.
Runs in a terminalAdded 2 Nov
agent-desktop Let AI agents read and click desktop apps
For developers building computer-use agents: agent-desktop lets an AI agent read and operate desktop apps on Mac, Windows or Linux from a command line.
+13,858 stars this weekAdded 3 Nov

Related projects

LFM3-Omni-2B Ask about photos, video and speech by voice (2.17B)
Same topic: transformers
Needs a model setupAdded 3 Nov
NVIDIA-Nemotron-3.5-Transcribe-30B-A4B Transcribe hour-long recordings (34B)
Same topic: transformers
Needs a model setupAdded 30 Oct
Mistral-Large-4-1T-A52B Chat, code and ask about images (1T, 52B active)
Same topic: transformers
Needs a model setupAdded 31 Oct
indra-ocr Read scanned pages (1.25M recognizer)
Same topic: safetensors
Runs in a terminalAdded 21 Oct
LFM3-1.7B Run local chat and tool-calling agents (1.73B)
Same topic: transformers
Needs a model setupAdded 28 Oct
drama-3 Generate multi-speaker audio (35B, 3B active)
Same topic: safetensors
Needs a model setupAdded 21 Oct

Explore by topic

Post this

For makers

Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.

README badge (Markdown)

[![Featured on BuildTube](https://buildtube.app/badge/hf-liquidai-lfm3-audio-2b.svg)](https://buildtube.app/p/hf-liquidai-lfm3-audio-2b/)

Embed card (HTML)

<iframe src="https://buildtube.app/embed/hf-liquidai-lfm3-audio-2b/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="LFM3-Audio-2B on BuildTube"></iframe>