Openlrc

Code refactoring, documentation updates, and minor bug fixes.

Other Changes:

- Fixed issue 54.
- Changed the default LLM model to `gpt-4o-mini`.
- Minor improvements on prompts.
- Improve post-optimizations for subtitles.

Minor bugfixes and improvements.

Other Changes:

- Fix default check_format returning False issue.
- Fix edges cases for srt file generation.
- Prepare translation evaluator for benchmarking.

This update add Gemini Models support for translation.

New Features:

- Support Gemini as translation engine.

Other Changes:

- Extract the `check_format` section into a standalone `validators.py` module.
- Add stop_sequences support for Chatbot.
- Use stop sequences when building Context.
- Also remove generated .wav files from videos if `clear_temp=True`.
- Minor improvements on Context Reviewer prompt.

Minor bugfixes and improvements.

Other Changes:

- Improve timestamp accuracy.
- Fix edges cases for transcription.

New Features:

- Add glossary support for domain specific translation.
- Add `retry_model` args for retrying translation with different models.
- Introduce `privider: model_name` for arbitrary model routing.

Other Changes:

- Extend 0.5s for each suitable sentence.
- Reduce noise suppression chunk-size for faster processing.
- Enhance translation workflow by Context Reviewer Agent.

New Features:

- Add custom endpoint (base_url) support for OpenAI & Anthropic.
- Generating bilingual subtitles.

Other Changes:

- Fix dep issues from ctranslate2 and streamlit-related packages.