naturalspeech3-facodec-v2 (Vokra GGUF)

Converted to the Vokra GGUF format for Vokra, a zero-dependency speech-AI inference runtime.

This is a conversion, not a new model. The weights are the upstream ones; Vokra re-packages them so its runtime can memory-map them directly. Credit for the model belongs upstream — see Source below.

Files

File	Size	SHA-256
naturalspeech3-facodec-v2.gguf	428.4 MB	ee1d1e23266d6d2a898152d18bde156a2de008b8fe1eae9eeb392feca24c3084

Usage

# Download (any HTTP client works — the file is a plain GGUF)
curl -L -o naturalspeech3-facodec-v2.gguf \
  https://huggingface.co/vokra/naturalspeech3-facodec-v2/resolve/main/naturalspeech3-facodec-v2.gguf

vokra-cli run --model naturalspeech3-facodec-v2.gguf --input input.wav

Provenance

Field	Value
Architecture	facodec
Tensors	806
Upstream source	amphion/naturalspeech3_facodec v2 (Amphion NaturalSpeech 3 FACodec — encoder_v2 + decoder_v2, factorized VQ codec, prosody 1cb + content 2cb + detail 3cb, 16 kHz, hop 200 → 80 tok/s, arXiv:2403.03100, apache-2.0)
Upstream licence	apache-2.0
Licence class	permissive
Registry model id	naturalspeech3-facodec-v2
Vokra GGUF schema	1
Converted by	vokra-core 0.1.0-alpha.0

Every row above is read out of this file's own vokra.* metadata, so the card cannot claim something the artifact does not carry.

Licence

The weights are distributed under apache-2.0, unchanged from upstream. Conversion does not alter the licence, and your obligations run to the upstream author.

Verifying this file

shasum -a 256 naturalspeech3-facodec-v2.gguf
# expect: ee1d1e23266d6d2a898152d18bde156a2de008b8fe1eae9eeb392feca24c3084