Hugging Face's logo

PraxySante
/
stt_ca-es_conformer_transducer_large_fine_tuned

Automatic Speech Recognition
NeMo
Spanish
Catalan
speech
multilingual
Eval Results
🇪🇺 Region: EU
Model card Files Community
stt_ca-es_conformer_transducer_large_fine_tuned
File size: 4,005 Bytes 
0d9e8fd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
 
---
datasets:
- facebook/multilingual_librispeech
- Parlament-Parla-v1
- gttsehu/basque_parliament_1
- facebook/voxpopuli
- johnatanebonilla/coser_lv_full
- collectivat/tv3_parla
- mozilla-foundation/common_voice_16_0
language:
- es
- ca
metrics:
- wer
- cer
tags:
- automatic-speech-recognition
- speech
- multilingual
- nemo
model-index:
  - name: Mohammed-Alzahrani-ai/stt_ca-es_conformer_transducer_large_fine_tuned
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          type: automatic-speech-recognition
          name: Combined (Parlament-Parla-v1, MLS, Voxpopuli, etc.)
        metrics:
          - name: WER (Spanish)
            type: wer
            value: 0.08
          - name: CER (Spanish)
            type: cer
            value: 0.04
          - name: WER (Catalan)
            type: wer
            value: 0.10
          - name: CER (Catalan)
            type: cer
            value: 0.05
---
# NVIDIA Conformer-Transducer Large (ca-es)

## Table of Contents
<details>
<summary>Click to expand</summary>

- [Model Description](#model-description)
- [Intended Uses and Limitations](#intended-uses-and-limitations)
- [How to Get Started with the Model](#how-to-get-started-with-the-model)
- [Training Details](#training-details)
- [Citation](#citation)
- [Additional Information](#additional-information)

</details>

## Summary

The "stt_ca-es_conformer_transducer_large" is an acoustic model based on ["NVIDIA/stt_es_conformer_transducer_large"](https://huggingface.co/nvidia/stt_es_conformer_transducer_large/) suitable for Bilingual Catalan-Spanish Automatic Speech Recognition.

## Model Description

This model transcribes speech, and was fine-tuned on a Bilingual ca-es dataset comprising of 4000 hours. It is a "large" variant of Conformer-Transducer, with around 120 million parameters. We expaneded it is tokenizer vocab sise to be 5.5k t oinclude lowercase, uppercase, and punctuation
See the [model architecture](#model-architecture) section and [NeMo documentation](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/main/asr/models.html#conformer-transducer) for complete architecture details.

## Intended Uses and Limitations

This model can be used for Automatic Speech Recognition (ASR) in Catalan and Spanish. It is intended to transcribe audio files in Catalan and Spanish to plain text with punctuation.

### Installation

To use this model, install [NVIDIA NeMo](https://github.com/NVIDIA/NeMo). We recommend you install it after you've installed the latest PyTorch version.
```
pip install nemo_toolkit['all']
``` 

### For Inference
To transcribe audio in Catalan or in Spanish using this model, you can follow this example:


```python
import nemo.collections.asr as nemo_asr
nemo_asr_model = nemo_asr.models.EncDecRNNTBPEModel.restore_from(model)
transcription = nemo_asr_model.transcribe([audio_path])[0].text
print(transcription)
```

## Training Details

### Training data

The model was fine-tuned on bilingual datasets in Catalan and Spanish, for a total of 4k hours. Including:
- [Parlament-Parla-v1](https://openslr.org/59/)
- [multilingual_librispeech](https://huggingface.co/datasets/facebook/multilingual_librispeech)
- [basque_parliament_1](https://huggingface.co/datasets/gttsehu/basque_parliament_1) 
- [Voxpopuli](https://huggingface.co/datasets/facebook/voxpopuli) (The datasets will be made accessible shortly.)
- [Coser](https://huggingface.co/datasets/johnatanebonilla/coser)
- [tv3_parla](https://huggingface.co/datasets/collectivat/tv3_parla)
- [common_voice_16_0](https://huggingface.co/datasets/mozilla-foundation/common_voice_16_0)

### Training procedure

This model is the result of finetuning the model ["projecte-aina/stt_ca-es_conformer_transducer_large"](https://huggingface.co/projecte-aina/stt_ca-es_conformer_transducer_large)



### Results

**Spanish WER:** 0.08
**Catalan WER:** 0.10

**Spanish CER:** 0.04
**Catalan CER:** 0.05