Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
TTS-Story is a web-based multi‑voice TTS studio for turning tagged scripts into audiobooks—featuring full speaker management, chunk review/regeneration, a job queue and library system, and local GPU or API backends including Kokoro, Chatterbox, VOX CPM, Pocket-TTS, Kitten-TTS, IndexTTS-2, QWEN3 TTS and Omnivoice engines
| Date | Stars |
|---|---|
| 2026-07-30 | 291 |
| 2026-08-07 | 307 |
| 2026-08-15 | 317 |
| 2026-08-18 | 318 |
| 2026-08-19 | 321 |
| 2026-08-21 | 322 |
| 2026-08-23 | 324 |
| 2026-08-30 | 325 |
| 2026-09-01 | 326 |
| 2026-09-02 | 325 |
| 2026-09-03 | 326 |
| 2026-09-04 | 327 |
| 2026-09-05 | 328 |
| 2026-09-06 | 329 |
| 2026-09-07 | 332 |
| 2026-09-08 | 333 |
| 2026-09-10 | 333 |
| 2026-09-11 | 334 |
| 2026-09-12 | 334 |
| 2026-09-13 | 334 |
| 2026-09-14 | 334 |
| 2026-09-15 | 335 |
| 2026-09-16 | 336 |
| 2026-09-17 | 336 |
| 2026-09-18 | 336 |
| 2026-09-21 | 337 |
| 2026-09-25 | 338 |
| 2026-09-29 | 337 |
| 2026-09-30 | 338 |
| 2026-10-02 | 339 |
| 2026-10-03 | 340 |
| 2026-10-04 | 340 |
| 2026-10-05 | 340 |
| 2026-10-06 | 341 |
| 2026-10-08 | 341 |
| 2026-10-12 | 341 |
| 2026-10-14 | 343 |
| 2026-10-15 | 342 |
| 2026-10-16 | 341 |
| 2026-10-19 | 342 |
| 2026-10-20 | 344 |
| 2026-10-24 | 345 |
| 2026-10-26 | 345 |
| 2026-10-27 | 345 |
| 2026-10-30 | 346 |
| 2026-10-31 | 348 |
| 2026-11-02 | 351 |
| 2026-11-04 | 351 |
Today
— stars today
This week
+6 stars this week
This month
+11 stars this month
Momentum
6.0
growth rate 1.74%/day
## Gemini TTS cloud preview Gemini candidates now appear directly beneath the existing Speaker Properties controls when the speaker opens, including candidates created in bulk. New auditions preserve the previously assigned voice until a replacement is selected. Pending auditions are retained without rebilling; completed/selected auditions can receive new candidates in a new batch. Bulk results distinguish newly created auditions from retained ones. **Generate Voice** uses that same profile's prompt, model, and candidate count; there is no second speaker-design form to open. Failed bulk requests are shown as errors instead of an unexplained empty audition, including Google's error reason when available. Cloud batch checkpoints save in parallel with the next speaker's requests (one save at a time), and pause/completion still wait for checkpoint persistence. **Bulk Gemini casting:** Use **Generate Voices** beside **Build Profiles**, select **Gemini TTS · Cloud Voice Design**, and choose Flash or Flash-Lite. Up to four candidates per speaker run concurrently; speakers run sequentially, with pause/resume checkpoints between speakers. Creation requires the dedicated Gemini TTS key and billing confirmation. Existing speaker auditions are preserved instead of generating duplicates. Completed candidates survive partial failures and are saved with the project; open each speaker to audition and select a voice (including single candidates). Google supplies the preview passage. Keep the page open during requests and save the project after selection. **Selecting a designed voice:** Gemini hosted voices require a Gemini narration engine. If another engine is selected, **Use voice** offers to switch the project's narration engine to Gemini before assigning the candidate. This affects all speakers, so review their assignments before generation. Cancelling preserves all candidates. After confirmed candidate cleanup, audition previews collapse and the selected voice remains in the speaker dropdown. In Speaker Properties, select **Gemini TTS · Cloud Voice Design** and click **Generate Voice**. Controls and compact candidate previews appear inline, using the speaker's description. Up to four candidates are requested concurrently (subject to Google quotas); successful candidates are retained even if another request fails. Requests are billable and creation is not automatically retried. Selecting a voice keeps the speaker window open; save your project afterward. The standalone designer remains available from Available Voices and settings. **Manage hosted custom voices:** Open **Available Voices → Gemini Custom Voices · Cloud** to refresh, search by name or ID, play previews, design new voices, or delete individual voices from Google. Compact rows also support checkbox selection, **Select visible**, **Clear selection**, and **Delete selected**. This includes older candidates with no audition grouping. Deletion requires confirmation and leaves existing audio files unchanged, but future synthesis/regeneration needs a replacement wherever the deleted ID was assigned. Preset voices cannot be deleted. Previews load only when Play is clicked and share one player. Renaming hosted voices is not currently offered. Gemini audition cleanup: after selecting a candidate, confirm removal of its unselected siblings from Google storage. Only IDs recorded together in that audition are removed; presets, the selected voice, and unrelated designs are preserved. Pending auditions belong to the speaker profile and are included when saving the project. Reopening Speaker Properties restores them; a project/speaker-scoped browser draft also protects completed candidates across page reloads. Save the project for portability to another browser. If none of the pending candidates suit the speaker, click **Generate Voice** again and confirm permanent deletion of that pending audition followed by billable replacement generation. Cancel keeps the candidates. Any deletion fa
Excerpt of 34,447 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:3f38728ef681be10, topic:tts, readme:text-to-speech, readme:voice cloning