Fix audio synchronization in VideoWriter multi-slot merging

[Copilot]

Audio from multiple video sources connected through ImageConcat was being incorrectly interleaved per-frame instead of concatenated per-slot, producing garbled output.

Root Cause

VideoWriter was merging audio chunks on each frame update:

# For each frame: sort all slots by timestamp, merge, append
Frame 1: [slot1_chunk, slot0_chunk] → merged_chunk1
Frame 2: [slot1_chunk, slot0_chunk] → merged_chunk2
Final: [merged_chunk1, merged_chunk2]  # Interleaved audio ❌

This breaks temporal coherence - each video source's audio should play sequentially, not interleaved.

Changes

VideoWriter data structure

• Changed _audio_samples_dict from list to dict[slot_idx, SlotData]
• SlotData = {'samples': list, 'timestamp': float, 'sample_rate': int}

Collection logic

• During recording: accumulate frames per-slot
• On recording stop: concatenate each slot's frames, then sort slots by timestamp

Result:

# Collect per slot
Slot 0 (ts=100.0): [frame1, frame2, ...] → concatenated
Slot 1 (ts=99.9):  [frame1, frame2, ...] → concatenated

# Sort by timestamp
Final: [slot1_audio, slot0_audio]  # Correct temporal order ✓

Testing

• Added test_video_writer_audio_slot_merge.py covering multi-slot, single-slot, and edge cases
• Existing audio sync tests continue to pass

Original prompt

le son ne marche pas lors de la fusion image son lors de ça ====> chunk de l'audio fusionne de façon synchronisé, quand on passe l'audio et la video au travzers de la syncQueue, tous cela au finale au traverse de imageconcat et videowriter. vérifie ça.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.