python_client: Avoid async in playback callback, use thread async for prebuffer, reduce blocksize and latency, keep minimal threshold and gain.

Revert TTS pause flush back to 600 ms as it is more natural, target NMT below this.

Pin TTS pause threshold with 1.2s for now, and revert max_workers to cpu_count.

Fix bug on log in python_client

Revise TTS pause threshold to 2 s. due to TranslateGemma bottleneck. Need to make NMT faster < 1 s.

Fix bugs on translation_merger and orchestrator. Revert workers to cpu_count + 4, and tts pause threshold.

Use log instead of print for python_client, remove audio stats print

Fix TTS queue with put_nowait, reduce pause threshold, thread to cpu count, and remove audio stats print.

Add the logs examples

Add examples for real-time server-client conn. and outputs

Add audio output file save on client too for checking

Fix NMT pause sync. with ASR threshold and thread counts for smoother processing

Update README

Revise python_client to handle proper playback FIFO buffering audio buffering. Still needs tweak for smoothing.

Fix is_punctuation_start on translation_merger and pause TTS threshold.

Remove commented function

Fix ASR-NMT and TTS on the server side. Still need to fix on the client side for playback.

Update README

Revise TTS queue with blocking

Fix for checking ASR-NMT output and sync. with TTS. TTS handling still needs more sync.

Update README command & sample rate

Add TTS model files

Add NMT model file

Add ASR model files

Change default client sample rate to 16k

Add client requirements

Formatting black, isort, flake8

Update README base model

Enhanced docstrings, type hints, and comments

Add project files

initial commit