---
license: mit
pipeline_tag: feature-extraction
tags:
  - dms
  - model-assets
base_model:
  - innnky/contentvec
  - safe-models/ContentVec
  - lj1995/VoiceConversionWebUI
---

<!-- dms-model-assets-publish-20261022T0750Z-v7 -->

# ContentVec 768L12

ContentVec feature encoder weights and configuration used by the local RVC pipeline.

## Files and local identity

- `config.json` — 1388 bytes; SHA-256 `2ddde063b795d38d9051a7215a092fecf4cfe148b54251e38de51d88d356898b`
- `pytorch_model.bin` — 378342945 bytes; SHA-256 `d8dd400e054ddf4e6be75dab5a2549db748cc99e756a097c496c099f65a4854e`

## Upstream copies and provenance

- Related upstream: [innnky/contentvec](https://huggingface.co/innnky/contentvec).
- Related upstream: [safe-models/ContentVec](https://huggingface.co/safe-models/ContentVec).
- Related upstream: [lj1995/VoiceConversionWebUI](https://huggingface.co/lj1995/VoiceConversionWebUI).
- Related upstream: [licyk/rvc-model](https://huggingface.co/licyk/rvc-model).
- No byte-identical copy for the remaining large local model files was found in the checked upstream trees: innnky/contentvec, safe-models/ContentVec, lj1995/VoiceConversionWebUI, licyk/rvc-model. These links are related sources, not proof of identical bytes.

## License and use

License metadata follows the identified upstream license `mit`. Upstream terms and attribution requirements apply; this repository does not claim ownership of upstream model weights.

## Provenance audit

Generated from local DMS model assets. Audit run: `dms-model-assets-publish-20261022T0750Z-v7`. SHA-256 values identify the local source bytes.