DialogueSidon — Two-Speaker Dialogue Separation & Restoration

Upload a degraded or noisy audio/video recording of a two-speaker conversation. DialogueSidon jointly separates the two speakers and restores clean, high-quality speech from the mixture — handling background noise, reverberation, and channel degradation in one pass. Inputs up to 120 s are processed in one shot. Longer inputs are processed in 120 s chunks with 10 s overlap crossfade and automatic speaker re-alignment across chunks. Model: sarulab-speech/DialogueSidon