Spaces:
Sleeping
Sleeping
| title: ASR model for Lesbian Greek | |
| emoji: 🚀 | |
| colorFrom: blue | |
| colorTo: purple | |
| sdk: gradio | |
| sdk_version: "5.35.0" | |
| app_file: app.py | |
| pinned: false | |
| license: mit | |
| models: | |
| - openai/whisper-medium | |
| - openai/whisper-large-v2 | |
| - openai/whisper-large-v3 | |
| - distil-whisper/distil-large-v2 | |
| - ilsp/whisper_greek_dialect_of_lesbos | |
| tags: | |
| - speech-to-text | |
| - whisper | |
| - transcription | |
| - greek | |
| - audio | |
| - asr | |
| - automatic-speech-recognition | |
| --- | |
| # ASR model for Lesbian Greek | |
| High-performance speech-to-text transcription using OpenAI Whisper with advanced optimizations. Features Flash Attention 2, anti-hallucination measures, and support for multiple models including Greek dialect specialization. | |
| ## Key Features | |
| - 🚀 **Flash Attention 2** for faster processing | |
| - 🛡️ **Anti-hallucination** optimizations | |
| - 🎯 **Multiple model support** including Greek dialect | |
| - ⚡ **Batch processing** for efficiency | |
| - 📊 **Real-time progress** tracking | |
| - 🔧 **Advanced configuration** options | |
| ## Recommended Settings | |
| - **Model**: `openai/whisper-medium` (best balance) | |
| - **Chunk Length**: 30 seconds | |
| - **Batch Size**: 16 | |
| - **Language**: Automatic Detection | |
| ## Performance | |
| - Up to **15x real-time** processing speed | |
| - **80%+ reduction** in hallucinations | |
| - Support for **multiple languages** | |
| - Optimized for **production use** |