Update README.md
Browse files
README.md
CHANGED
@@ -44,14 +44,14 @@ model-index:
|
|
44 |
config: Ukrainian P&C
|
45 |
split: test
|
46 |
args:
|
47 |
-
language:
|
48 |
metrics:
|
49 |
- name: Test WER P&C
|
50 |
type: wer
|
51 |
value: 6.92
|
52 |
---
|
53 |
|
54 |
-
# NVIDIA FastConformer-Hybrid Large (
|
55 |
|
56 |
<style>
|
57 |
img {
|
@@ -83,7 +83,7 @@ The model is available for use in the NeMo toolkit [3], and can be used as a pre
|
|
83 |
|
84 |
```python
|
85 |
import nemo.collections.asr as nemo_asr
|
86 |
-
asr_model = nemo_asr.models.EncDecHybridRNNTCTCBPEModel.from_pretrained(model_name="nvidia/
|
87 |
```
|
88 |
|
89 |
### Transcribing using Python
|
@@ -101,14 +101,14 @@ asr_model.transcribe(['2086-149220-0033.wav'])
|
|
101 |
Using Transducer mode inference:
|
102 |
```shell
|
103 |
python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py
|
104 |
-
pretrained_name="nvidia/
|
105 |
audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
|
106 |
```
|
107 |
|
108 |
Using CTC mode inference:
|
109 |
```shell
|
110 |
python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py
|
111 |
-
pretrained_name="nvidia/
|
112 |
audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
|
113 |
decoder_type="ctc"
|
114 |
```
|
|
|
44 |
config: Ukrainian P&C
|
45 |
split: test
|
46 |
args:
|
47 |
+
language: ua
|
48 |
metrics:
|
49 |
- name: Test WER P&C
|
50 |
type: wer
|
51 |
value: 6.92
|
52 |
---
|
53 |
|
54 |
+
# NVIDIA FastConformer-Hybrid Large (ua)
|
55 |
|
56 |
<style>
|
57 |
img {
|
|
|
83 |
|
84 |
```python
|
85 |
import nemo.collections.asr as nemo_asr
|
86 |
+
asr_model = nemo_asr.models.EncDecHybridRNNTCTCBPEModel.from_pretrained(model_name="nvidia/stt_ua_fastconformer_hybrid_large_pc")
|
87 |
```
|
88 |
|
89 |
### Transcribing using Python
|
|
|
101 |
Using Transducer mode inference:
|
102 |
```shell
|
103 |
python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py
|
104 |
+
pretrained_name="nvidia/stt_ua_fastconformer_hybrid_large_pc"
|
105 |
audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
|
106 |
```
|
107 |
|
108 |
Using CTC mode inference:
|
109 |
```shell
|
110 |
python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py
|
111 |
+
pretrained_name="nvidia/stt_ua_fastconformer_hybrid_large_pc"
|
112 |
audio_dir="<DIRECTORY CONTAINING AUDIO FILES>"
|
113 |
decoder_type="ctc"
|
114 |
```
|