Speech-to-text prototype
Multi-model · CPU ·
no model loaded
🎙️ Record
0:00
ready
🎲 Random Polish sample
ground truth
playback:
Model:
Parakeet TDT 0.6B v3 (NVIDIA)
Qwen3-ASR 0.6B
Qwen3-ASR 1.7B
Transcribe
loading model
running inference
done