Repository navigation
v2.3.0 Release
JP
概要
CapCut の最新バージョン対応と音声カタログ取得の刷新、Docker イメージ公開環境を追加したリリースです。
主な変更点
- CapCut の最新バージョンへ対応(editor 署名の appvr を 5.8.0 へ更新、バージョン系設定は live bundle から自動取得)
- 音声カタログ取得を
get_panel_info一括取得を第一経路に変更(カテゴリを動的に解決、失敗時は従来のカテゴリ別取得へフォールバック) - TTS 非対応の項目(
effect_typeが 11 以外)を話者一覧から除外 isVipフィールドとCAPCUT_EXCLUDE_VIP_VOICES(VIP かつ 11labs 連携音声の除外設定)を追加- legacy 署名をリクエスト時に自動生成するように変更し、
LEGACY_DEVICE_TIME/LEGACY_SIGNを廃止 - GHCR への Docker イメージ公開 workflow を追加(main >
latest+ バージョンタグ、develop >nightly)とdocker-compose.yamlを同梱 - bundle 抽出と
CapCutServiceのリファクタ(未使用コード・環境変数の削除)
備考
- 既存の
/v2/synthesizeの使い方はそのままです。レスポンスの変更は話者一覧のisVip追加のみです .envにLEGACY_DEVICE_TIME/LEGACY_SIGNが残っていても無視されますdocker compose up -dでは GHCR イメージを使用し、書き込み状態はdata/へ保存されます
EN
Summary
Release with support for the latest CapCut version, a renewed voice catalog flow, and a Docker image publishing setup.
Highlights
- Support the latest CapCut version (editor sign
appvrupdated to 5.8.0, version settings are now fetched from live bundles) - Voice catalog now uses a single
get_panel_inforequest as the primary route (dynamic category resolution, falling back to per-category requests) - Exclude non-TTS entries (
effect_typeother than 11) from the speaker list - Added
isVipfield andCAPCUT_EXCLUDE_VIP_VOICES(excludes VIP voices backed by the 11labs integration) - Legacy signatures are now generated per request;
LEGACY_DEVICE_TIME/LEGACY_SIGNare no longer needed - Added a GHCR Docker image workflow (main >
latest+ version tag, develop >nightly) withdocker-compose.yaml - Refactored bundle extraction and
CapCutService(removed unused code and env vars)
Notes
- Existing
/v2/synthesizeusage stays the same. The only response change is the newisVipfield in the speaker list LEGACY_DEVICE_TIME/LEGACY_SIGNin.envare ignored if presentdocker compose up -duses the GHCR image and stores writable state indata/