Instructions to use mlx-community/nemo-nano-codec-22khz-1.89kbps-21.5fps with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/nemo-nano-codec-22khz-1.89kbps-21.5fps with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir nemo-nano-codec-22khz-1.89kbps-21.5fps mlx-community/nemo-nano-codec-22khz-1.89kbps-21.5fps
- NeMo
How to use mlx-community/nemo-nano-codec-22khz-1.89kbps-21.5fps with NeMo:
# tag did not correspond to a valid NeMo domain.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
NeMo NanoCodec 22 kHz / 1.89 kbps / 21.5 fps — MLX (safetensors)
MLX-ready safetensors conversion of NVIDIA's NeMo NanoCodec
(nvidia/nemo-nano-codec-22khz-1.89kbps-21.5fps),
the 22.05 kHz / 21.5 fps / 1.89 kbps variant. It is a neural audio codec using finite scalar
quantization (GroupFSQ, 8 groups × FSQ levels [8, 7, 6, 6], 32 embed dims) with a HiFi-GAN
encoder/decoder.
This repo exists so Swift-MLX / MLX consumers can load the codec weights directly — the
upstream distribution ships a .nemo archive that requires the NeMo/PyTorch stack (Linux-only in
practice). The tensors here are extracted from that archive verbatim (weight-norm kept in the
modern parametrizations.weight.{original0,original1} layout; no numerics changed). It is the
codec layer of the Gepard-1.0 streaming-TTS port
(xocialize/mlx-gepard-swift).
Contents
| file | what |
|---|---|
model.safetensors |
encoder (audio_encoder.*) + decoder (audio_decoder.*) + quantizer weights (fp32) |
codec_config.yaml |
the NeMo config the weights were exported with (rates, dims, FSQ levels) |
nvidia-open-model-license-agreement-june-2024.pdf |
the governing license (shipped per §3.1) |
NOTICE |
the §3.1 attribution notice |
Parity
Converted and validated against the NeMo reference (CPU fp32) as part of the Gepard port:
- decode (32-ch codes → waveform): max|Δ| 6.7e-6, SI-SNR 104 dB, corr 1.000000
- encode (waveform → codes): 99.5% packed integer-exact (residual = FSQ knife-edge boundary flips + ≤1 tail frame — the documented conv-edge convention).
License / Terms of Use
Governed by the NVIDIA Open Model License Agreement (v. June 14, 2024) — see the bundled
PDF and NOTICE. The model is commercially usable; redistribution of the model and derivatives
is permitted with the §3.1 obligations (ship a copy of the Agreement and the Notice below).
Licensed by NVIDIA Corporation under the NVIDIA Open Model License
The conversion tooling and the consuming Swift port are Apache-2.0; only these weights carry the NVIDIA Open Model License.
Citation
NeMo NanoCodec — https://arxiv.org/abs/2508.05835
Quantized