Name the voice files after the voices

Bump serverless image tag to v7

Withdraw the Maya voice

Ignore sparse nikud, and split partial nikud between front ends

Treat Russian hyphenated compounds as one word, not two

Make the GPU the default on ZeroGPU

Give ORT's CUDA provider sane options, and time inside the GPU window

Let each PT checkpoint come from its own repo and path

Accept v2.5 PT checkpoints and verify them against the bundle by hash

Fix Russian hyphens and clause dashes

Default ZeroGPU back to the CPU session, now that the GPU path is measurable

Detect ZeroGPU by truthiness, not the literal "1"

Split the CPU-onnxruntime dependencies into pre-requirements.txt

Install renikud-plus from the PR branch so ONNX gets the GPU

Report g2p time in the stats bar

Memoize RUPhon's per-word inference

Stop the Hebrew and Russian front ends oversubscribing the CPU

Close the gap between synthesized chunks

Correct the onnxruntime notes: ordering does not fix the collision

Install onnxruntime last so the GPU wheel is not clobbered

Make ORT find CUDA on ZeroGPU, and let the audio player rewind

Only enter the ZeroGPU window when it actually yields a GPU

Prewarm the Russian front end instead of stalling the first request

Load the Russian front end outside the ZeroGPU window

Move to BlueTTS 2.5: new ONNX bundle, named voices, Russian, RenikudPlus

Keep synthesis on ZeroGPU only — retry instead of CPU fallback

Fall back to ONNX CPU when ZeroGPU hits CUDA ECC faults

Fix stale audio by disabling Gradio streaming replace mode

Reset Gradio streaming audio at start of each generation

Fix ZeroGPU synth: do not re-import app inside PT worker

Use PyTorch Hub weights for real ZeroGPU CUDA synthesis

Run BlueTTS on Hugging Face ZeroGPU

Stream synthesis chunks to Gradio Audio as they finish

Improve Hebrew text preprocessing for the Space

DP ref encoder uses compressed latent channels from config; README v2 onnx path; pytest smoke tests

Hebrew dates: spell month with ordinal names (לראשון … לשנים עשר)

Hebrew: format expanded dates as 'D ל-M Y'

Expand DD/MM/YYYY-style dates before generic number expansion

Hebrew: expand 'ל' before Latin tokens (e.g. CPU-GPU) to 'אל'

Lower BLUE_SYNTH_MAX_CHUNK_LEN to 200

Add phonemes_for_display and strip language tags for user-facing text

Strip Hebrew gershayim/quote marks in abbreviations before synthesis

Hebrew+Latin split phonemization; email/%/ratio/number prep; per-chunk phonemize; mixed pace blend

Tuning: CFG 4.0; tighten steps (5-16) and speed (0.8-1.2) slider ranges

Strip Hebrew nikud; normalize 'any more'; default speed 0.95, CFG 3.5

Download default female/male v2 voice JSONs from hub; restore style_ttl std filter; remove Refresh voices UI

Voices: require style_ttl+style_dp; add Refresh voices button; select new clone in dropdown

Gradio: drop CFG/pace sliders (use cfg 3, pace blend 0); fix bad indents in infer loop and helpers

Clone export: reference audio cleanup and left-aligned ref mask; PT weights bundle marker and force download

Strip XML-like tags with attributes; unify tag stripping for phonemize and char encode