Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

feat: one-click gallery of curated DisfluencySpeech demo clips under the recorder

add curated DisfluencySpeech demo clips

fix: diff compares words only (ignore punctuation/casing), strike true deletions only — removes false-positive strikethrough noise

fix: remove startup model preload that crashed ZeroGPU (No CUDA GPUs at import); load lazily inside @spaces.GPU transcribe

feat: expand + humanize facts carousel (8 facts, no em dashes), fix stale docstring and waveform comment

Upload folder using huggingface_hub

Upload folder using huggingface_hub

docs: credit LibriSpeech+Kokoro synthetic training data, clarify DisfluencySpeech is the eval benchmark, humanize copy, fix hardware to zero-a10g

fix: native audio player (kills mid-record freeze + 1x overflow), single base+LoRA load, compact hero with CTA, humanized copy

Upload folder using huggingface_hub

Upload folder using huggingface_hub

readme: add submission tags, Modal training note, marketing copy

ui: full-bleed cream layout (no dark bars), warm theme kills indigo chrome, flex hero copy, no em dashes

space card: coffee theme (emoji+colors)

redesign: premium coffee-and-cream espresso UI; drop examples+caution; add footer

Upload folder using huggingface_hub

initial commit