Add: Conette version, paper link and github link on the bottom of the page.

Mod: Format file using black and update help msg.

Mod: Rework UI, remove tmp files and clear cache after 10min.

Mod: Update error msg when audio file is too short or too long.

Add/Fix: Add spinner during generation and fix results order.

Mod: Update UI to store microphone input in microphone_conette_record.wav file, raises an error when the audio is too short or too long, update main description and show other candidates in outputs.

Add/Mod: Add new audio recorder, threshold hparam, min and max beam size single slider.

Mod: Update typing.

Fix: Avoid runtime error when candidate is empty.

Mod: Refactor model forward.

Mod: Update warning msg.

Add: Recording option for testing CoNeTTE.

Mod: Update forward to compute all audio files per batch and improve UI for hyperparameters.

Add: Allow repetition mode option.

Mod: Cache candidate results in memory to avoid re-computation.

Mod: Update hparams ranges.

Add: Options beam size, min pred size and max pred size to UI.

Add: CoNeTTE streamlit application for inference on files.

Add: Application file.