NVIDIA NemotronLabs RE-USE-laughter GGUF for speech.cpp

This file runs only in speech.cpp 0.10.1 or later. Its layout is speech.cpp's own, so llama.cpp, LM Studio, Ollama and other GGUF readers cannot run it.

It holds NVIDIA NemotronLabs RE-USE-laughter, a SEMamba network of 9.6M parameters in F32, its STFT's settings and the chunks of the official inference_chunk.py. It takes recorded speech in any language at any rate from 8 to 48 kHz and gives back the same recording, at the same rate, for people to listen to. It removes noise, reverberation, clipping, band limits and codec damage as RE-USE does, and keeps laughter, which RE-USE removes with the noise.

Use

Install speech.cpp, then:

speech enhance re-use-laughter noisy.wav -o clean.wav
speech enhance re-use-laughter --bwe 48000 phone-call.wav -o wide.wav      # resampled to 48 kHz first, to extend its band
speech asr qwen3-asr-1.7b --enhance re-use-laughter crowd.wav              # each file enhanced whole before it is recognized
speech serve re-use-laughter                                               # POST /v1/audio/enhancements

The name re-use-laughter fetches this file the first time it is used; a path to the downloaded file works as well. A program calls the C API's speech_enhance(), or its enhancements of audio given a piece at a time. speech.cpp's documentation has the model's page, the command line and the server.

Files

Accuracy

Each stage is checked against the official code at commit 31f82f2 on FLEURS speech under babble at 16 kHz, in chunks of 5 s: on an Apple M5 the network lies 75.9 to 78.7 dB from the official's on each chunk, on the CPU and on Metal, and the enhanced audio 63.4 dB.

Speed

The RTFx (the seconds of audio over the seconds they took) of speech enhance on 60 s of noisy speech, the median of three runs and one on the CPU, 2026-11-01:

Each part of the audio is heard twice, as a chunk of 5 s starts every half chunk; an utterance enhanced whole, as speech asr --enhance takes it, is heard once.

Source

Converted by speech.cpp's reference/semamba/convert.py from model.safetensors (38,592,940 bytes, SHA-256 35c2a4e284515275d2e49a8e8cb5cd392ae67a1f1d345de1c12a0202c7b6b610) of nvidia/NVIDIA-NemotronLabs-RE-USE-laughter at revision 1b1521dd659608e1e063ada83e0acfe1da473a8b, with the settings of its recipe and its inference_chunk.py, from the code of nvidia/NVIDIA-NemotronLabs-RE-USE at revision 31f82f2efad1095034492f236e8dac9239ba713b. Its sibling, re-use, removes laughter with the noise.

License

NVIDIA NemotronLabs RE-USE-laughter is NVIDIA's, released under the OpenMDW License Agreement, version 1.1, which LICENSE.OpenMDW-1.1 holds. This file is distributed under that agreement. NVIDIA's model card has the model's ethical considerations.