Spaces:
Running
Running
Commit History
14 days ago
-
Edge engine: streaming with Matcha-TTS-PL-Edge (chunk / ODE / vocoder context controls); sentences merged to >= 40 chars
Verified signature528265b
22 days ago
-
No voice memory: every visit starts with the first (best) voice
Verified signature7d07ac4 -
Voice switching: one model in memory at a time (previous session released), loads serialized, models cached in the browser; defaults T 0.8 / tempo 0.8
Verified signature6dc1339 -
Real-time mode: one sentence per chunk
Verified signatureeea1f5d -
Apply the per-voice high-pass post-filter (female_voxcpm_1: 90 Hz), hidden from the UI
Verified signature1a982d2
23 days ago
-
Model renamed to Matcha-TTS-PL (base: Matcha-TTS-PL-Pretrain); four voices in the card
Verified signature2fdc2d9
24 days ago
-
Voice row: remove the leftover single-voice preset chip
Verified signature54db59f -
Voice selector: four synthetic voices, one model + own vocoder each (onnx/voices_index.json)
Verified signatureb9a4826
29 days ago
-
Examples: HVSR intro first, humanoid examples without questions
Verified signatureb1fd63e -
English words: phrase matching, Polish words (SGJP) read as Polish unless listed or in braces; switch to turn English detection off
Verified signature5b72baf -
Questions: '?' glued to the word, as in v2's training text (a space made wh-questions rise)
Verified signature2125ef1 -
Default temperature 1.0
Verified signature15b353a -
Always load the latest scripts (versioned imports, service worker revalidates same-origin files)
Verified signature3132ed5 -
Voice renamed to male_voxcpm_1; space before '?' for question intonation
Verified signature5cf08b3
-
Matcha-TTS-PL v2 browser playground
Verified signature735f5c2 -
initial commit
Verified signature37032b5