mARTin-B78/tts-dgx-spark-faster-qwen3-tts
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: add v6.2 changelog entries for streaming fix and voice field persistence

Browse Source 
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
mARTin-B78 2026-05-30 21:05:43 +02:00
parent 08be676685
commit 4a199a883a
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
README.md
View File

@ -336,6 +336,12 @@ The first request after container startup can be slower because CUDA graph captu
## Changelog ## Changelog
### v6.2 — 2026-05-30
**Fix: voice drift on streaming port 8023 + per-voice temperature persists across restarts**
- Streaming service (port 8023) was using the pre-v5 image (`martinb78/faster-qwen3-tts-dgx-spark:streaming`) which still had `non_streaming_mode=False` — the same voice drift bug as voiceclone had before v5. Fixed by switching streaming to `:latest` which carries the v5 patch. No image rebuild needed; compose-only change.
- `config/generate_voices.py` previously overwrote `voices.json` completely on every container start, discarding any manually added `temperature`, `top_k`, or `top_p` fields. Now merges from the existing `voices.json` so user-added sampling parameters survive restarts.
### v6 — 2026-05-30 ### v6 — 2026-05-30
**Restore `--max-seq-len`, consolidate Docker files, merge streaming repo** **Restore `--max-seq-len`, consolidate Docker files, merge streaming repo**