🎙️ Voice Agent Modular Testing Suite
Test individual components or the full voice agent pipeline. Each tab allows you to:
- STT Tab: Test speech-to-text models independently
- TTS Tab: Test text-to-speech models independently
- LLM Tab: Test language models independently
- Pipeline Tab: Test the complete voice agent flow (STT → LLM → TTS)
🎤 Speech-to-Text Testing
Select STT Model
Test Transcription
🔊 Text-to-Speech Testing
Select TTS Model
Test Synthesis
🤖 LLM Testing
Select LLM Model
Chat Interface
🎙️ Full Voice Agent Pipeline
Test the complete flow: Voice Input → STT → LLM → TTS → Voice Output
1. Load Models
STT Model
LLM Model
TTS Model
2. Voice Conversation
Conversation History
📝 Usage Tips
- Load models first: Click "Load Model" buttons before testing
- Recording audio: Click the microphone icon 🎤 to start recording, click again to stop
- Upload audio: Or drag & drop an audio file
- GPU acceleration: Models run on GPU if available, otherwise CPU
- Pipeline mode: Combines all modules for end-to-end voice interaction
- Performance: Use smaller models (Whisper Base, Qwen 0.5B) for faster performance on CPU