VOICEVOX CORE (VOICEVOX/voicevox_core): embeddable speech synthesis library (C API, Python wheel)
Source: https://github.com/voicevox/voicevox_core
MIT code with separately licensed voice models is the main license trap; compare voicevox--voicevox-engine (LGPL v3 dual).
Structured fields
{
"archived": false,
"revision": {
"ref": "0.17.0 (latest release, published 2026-08-13)",
"pinned": true,
"source_url": "https://github.com/VOICEVOX/voicevox_core/releases"
},
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "The MIT-licensed synthesis core behind VOICEVOX, shipped as dynamic libraries and Python wheels with a downloader for ONNX Runtime and voice-model (VVM) files. Lets an agent embed VOICEVOX without running the HTTP engine; the voice models it downloads have their own terms in a README inside the download.",
"source_type": "repository",
"availability": "available",
"dependencies": [
"VOICEVOX ONNX Runtime (fetched by the Downloader)",
"voice models (VVM files) fetched by the Downloader",
"Python 3 for the wheel"
],
"source_owner": "VOICEVOX project (Hiroshiba Kazuyuki)",
"next_check_at": "2026-10-18T00:00:00.000Z",
"hardware_notes": "Downloader flags select CPU, DirectML (--devices directml) or CUDA (--devices cuda) builds; no VRAM figures given in the guide pages read.",
"known_limitations": [
"The README warns that prebuilt core libraries in releases below version 0.16 are under a different license; only 0.16 and later are covered by MIT.",
"The MIT license covers the core source and build artifacts, not the voice models (VVM) which have their own terms.",
"Only the README, LICENSE and the user-guide page were read; per-language API docs were not."
],
"upstream_updated_at": "2026-10-17T17:48:12.000Z"
}Evidence claims
- [sourced_fact] The README states the source code and build artifacts of VOICEVOX CORE are under the MIT license, but prebuilt core libraries in releases below version 0.16 have a different license. · locator: README, ライセンス source
- [sourced_fact] Releases carry prebuilt C API dynamic libraries (.so/.dll/.dylib) and a Python API wheel. · locator: README intro source
- [sourced_fact] Running the core needs VOICEVOX ONNX Runtime and voice models (VVM files), which the core Downloader fetches; the guide notes the voice models have terms of use documented in a README inside the download. · locator: docs/guide/user/usage.md, 環境構築 source
- [sourced_fact] The LICENSE file is the MIT License with copyright to Hiroshiba Kazuyuki (2021). · locator: LICENSE source
- [sourced_fact] GitHub API returns the repository record for VOICEVOX/voicevox_core: default branch main; pushed_at at check time 2026-10-17T17:48:12Z (raw response saved under evidence/github/voicevox--voicevox-core.json). · locator: GitHub API repos/VOICEVOX/voicevox_core, default_branch (pushed_at is volatile) source
- [sourced_fact] GitHub API reports archived = false for VOICEVOX/voicevox_core. · locator: GitHub API repos/VOICEVOX/voicevox_core, archived source
- [sourced_fact] GitHub license detection for VOICEVOX/voicevox_core returns spdx_id MIT. · locator: GitHub API repos/VOICEVOX/voicevox_core, license source
- [sourced_fact] GitHub API latest release for VOICEVOX/voicevox_core: tag 0.17.0 published 2026-08-13T20:17:15Z (selected fields of the response saved under evidence/github/voicevox--voicevox-core.release.json). · locator: GitHub API repos/VOICEVOX/voicevox_core/releases/latest, tag_name source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| code | MIT | permitted | permitted | permitted | unknown | MIT for source and build artifacts of 0.16 and later; earlier prebuilt libraries have a different license. Notice retention required. Does not address training. |
| model_weights | unknown | unknown | unknown | unknown | unknown | Voice models (VVM) are separate from the MIT code and have their own terms in a README inside the downloaded files; that README was not read. |
| other | unknown | unknown | unknown | unknown | unknown | outputs: generated audio terms come from the voice-model terms and the VOICEVOX software terms (credit required); not stated in the core README. |
Linked records
- Do VOICEVOX voice-library terms (beyond the zunko.jp family) allow or restrict using generated audio as training data for other speech models? discusses → this
- Method (not legal advice). 1. Separate the components: VOICEVOX engine and editor code are dual-licensed LGPL v3 or a separate license, the core library is MIT for 0.16 and later (earlier prebuilt cores are under a different license), and the voice models (VVM) have their own terms in a README insi… cites → this
- No. GitHub reports NOASSERTION for the engine and editor (README: dual license, LGPL v3 or a separate license that does not require source disclosure) and MIT for the core library. The core README adds that prebuilt core libraries in releases below version 0.16 are under another license. The voice … cites → this