Model
NVIDIA
STT Pl FastConformer Hybrid Transducer-CTC Large P&CThis collection contains the large version (114M) of the Polish speech recognition model with a FastConformer encoder and a Hybrid decoder (joint RNNT-CTC loss). The model has a vocab size of 1024 and emits text with punctuation and capitalization.
Use the NGC CLI to download:
Copied!
2 Versions
1.21.0Selected07/20/2023 11:33 PM UTC437.96 MB Copied!
1.21.0Selected
07/20/2023 11:33 PM UTC437.96 MB
Copied!
Accuracy
| Key | Value |
|---|---|
| MCV12.0 TEST | 8.3 % |
| MLS DEV | 6.8 % |
| MCV12.0 DEV | 5.3 % |
| VOXPOPULI DEV | 10.9 % |
| VOXPOPULI TEST | 8.1 % |
| MLS TEST | 5.1 % |
Accuracy-PC
| Key | Value |
|---|---|
| MCV12.0 TEST | 10.6 % |
| MLS DEV | 15.0 % |
| MCV12.0 DEV | 7.9 % |
| VOXPOPULI DEV | 13.6 % |
| VOXPOPULI TEST | 11.0 % |
| MLS TEST | 9.8 % |
Model
| Key | Value |
|---|---|
| ARCHITECTURE | FastConformer Hybrid Transducer-CTC Large |
| INPUTS | 16000 KHZ MONO-CHANNEL AUDIO (WAV FILES) |
| OUTPUTS | TRANSCRIBED SPEECH |
05/02/2023 6:51 PM UTC437.98 MB Copied!
05/02/2023 6:51 PM UTC437.98 MB
Copied!
Accuracy
| Key | Value |
|---|---|
| MCV12.0 TEST | 8.7 % |
| MLS DEV | 7.1 % |
| MCV12.0 DEV | 6.0 % |
| VOXPOPULI DEV | 11.3 % |
| VOXPOPULI TEST | 8.5 % |
| MLS TEST | 5.8 % |
Accuracy-PC
| Key | Value |
|---|---|
| MCV12.0 TEST | 11.0 % |
| MLS DEV | 16.0 % |
| MCV12.0 DEV | 8.9 % |
| VOXPOPULI DEV | 14.0 % |
| VOXPOPULI TEST | 11.4 % |
| MLS TEST | 11.0 % |
Model
| Key | Value |
|---|---|
| ARCHITECTURE | FastConformer Hybrid Transducer-CTC Large |
| INPUTS | 16000 KHZ MONO-CHANNEL AUDIO (WAV FILES) |
| OUTPUTS | TRANSCRIBED SPEECH |