Model reference · open weights
jhcodec is an open-weight audio or speech model from jhcodec, listed in the AxForge catalogue. AxForge can bring it up on EU-owned hardware for you on request — with the licence handled where one is required.
About
Model Card for JHCodec JHCodec is a pure Transformer decoder-based neural audio codec with residual vector quantization (RVQ). It achieves state-of-the-art performance with minimal latency and high intelligibility through self-supervised representation reconstruction (SSRR) loss. - Paper: Reconstruct! Don't Encode: Self-Supervised Representation Reconstruction Loss for High-Intelligibility and Low-Latency Streaming Neural Audio Codec - GitHub Repository: https://github.com/jhcodec843/jhcodec - Demo: https://jhcodec843.github.io/jhcodec/ - License: MIT Model Details This checkpoint corresponds to the JHCodec-1M model variant (jhcodecmimi1000000.pt). JHCodec uses a self-supervised representation reconstruction loss to improve codec training, enhancing intelligibility by reconstructing distilled self-supervised representations from codec outputs. It features a zero-lookahead architecture designed for real-time streaming deployment. The model operates on 16 kHz mono audio in frames of FRAMESIZE = 320 samples (20 ms), so the input length must be a multiple of 320. Requirements - Python = 3.10 - PyTorch/TorchAudio with CUDA support (tested with torch==2.6.0+cu124 and torch==2.9.1+cu128) - omegaconf==2.3.0 - Flash-Attention (required for the reported performance; tested with flash-attn==2.7.4.post1 and flash-attn==2.8.3) - huggingfacehub — only if you auto-download the official checkpoint Note: Running on CPU currently leads to degraded reconstruction quality. Usage Inference via CLI Auto-download this checkpoint from the Hub: Or use a local copy of the checkpoint: Use in Python (offline, whole utterance at once) Use in Python (streaming, frame by frame) Pass the returned inferencecache back in on every call. The encoder and the decoder each keep their own cache, so use two separate variables and start both at None. To load a local checkpoint instead of the Hugging Face one: For CUDA-graph per-frame streaming (JHCodecMimiCudaGraph, whose statedict is identical to JHCodecMimi), see the GitHub repository README. Intended Use - Real-time low-latency audio codecs for speech-to-speech models - Research into neural codecs and generative modeling - Serving as a neural front-e
Summarised from the published model card. Read the full card on the HuggingFace links below.
Specifications
| Maker | jhcodec |
|---|---|
| Type | Audio & music |
| Variants | 1 |
| Released | 2026-03-08 |
| Popularity | 2k downloads / month |
| Likes | 4 |
| Licence | Open weights |
How it works
Variants
Open weights ship in several sizes and precisions. One page, all the variants — pick the one that fits your GPU. VRAM figures are estimates from model size.
| Variant | Params | Precision | VRAM | Fits 16 GB | Weights |
|---|---|---|---|---|---|
| jhcodec | — | BF16 | — | — | Weights ↗ |
Using it via the API
Once AxForge deploys jhcodec for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (jhcodec below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/audio/transcriptions \ -H "Authorization: Bearer $AXFORGE_API_KEY" \ -F model="jhcodec" -F file=@audio.mp3
Licence
Open weights under mit — commercial use is permitted. Deploy it on AxForge EU hardware on request. Read the licence ↗