Instructions to use SpireLab/spire_respin_baselines_espnet with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- ESPnet
How to use SpireLab/spire_respin_baselines_espnet with ESPnet:
from espnet2.bin.asr_inference import Speech2Text model = Speech2Text.from_pretrained( "SpireLab/spire_respin_baselines_espnet" ) speech, rate = soundfile.read("speech.wav") text, *_ = model(speech)[0] - Notebooks
- Google Colab
- Kaggle
Update consolidated README and meta.yaml with results
Browse files
README.md
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# ESPnet ASR Models in 9 Indian languages trained on RESPIN-S1.0 Small Train sets
|
| 2 |
+
|
| 3 |
+
This repository contains E-Branchformer-based ESPnet2 ASR models trained on the RESPIN-S1.0 small train splits.
|
| 4 |
+
|
| 5 |
+
## 🛠 Demo: How to use in ESPnet2
|
| 6 |
+
|
| 7 |
+
Follow the [ESPnet installation instructions](https://espnet.github.io/espnet/installation.html) if you haven't done that already.
|
| 8 |
+
|
| 9 |
+
```bash
|
| 10 |
+
cd espnet
|
| 11 |
+
pip install -e .
|
| 12 |
+
cd egs2/respin_small/asr1
|
| 13 |
+
./run.sh --skip_data_prep true --skip_train true --download_model SpireLab/spire_respin_baselines_espnet
|
| 14 |
+
```
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
## 📊 Results (CER/WER from RESULTS.md)
|
| 18 |
+
|
| 19 |
+
| Language | Model Name | CER (%) | WER (%) |
|
| 20 |
+
|----------|------------|---------|---------|
|
| 21 |
+
| bh | exp_small/exp_bh/asr_bh_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 4.4 | 15.2 |
|
| 22 |
+
| bn | exp_small/exp_bn/asr_bn_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 4.1 | 15.0 |
|
| 23 |
+
| ch | exp_small/exp_ch/asr_ch_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 3.1 | 10.6 |
|
| 24 |
+
| hi | exp_small/exp_hi/asr_hi_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 3.1 | 9.9 |
|
| 25 |
+
| kn | exp_small/exp_kn/asr_kn_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 4.6 | 24.5 |
|
| 26 |
+
| mg | exp_small/exp_mg/asr_mg_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 6.0 | 20.4 |
|
| 27 |
+
| mr | exp_small/exp_mr/asr_mr_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 3.1 | 14.5 |
|
| 28 |
+
| mt | exp_small/exp_mt/asr_mt_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 5.0 | 17.9 |
|
| 29 |
+
| te | exp_small/exp_te/asr_te_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1 | 4.1 | 21.6 |
|
| 30 |
+
|
| 31 |
+
---
|
| 32 |
+
|
| 33 |
+
## 📚 Citation
|
| 34 |
+
|
| 35 |
+
```bibtex
|
| 36 |
+
@inproceedings{watanabe2018espnet,
|
| 37 |
+
author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
|
| 38 |
+
title={{ESPnet}: End-to-End Speech Processing Toolkit},
|
| 39 |
+
year={2018},
|
| 40 |
+
booktitle={Proc. Interspeech},
|
| 41 |
+
pages={2207--2211},
|
| 42 |
+
doi={10.21437/Interspeech.2018-1456},
|
| 43 |
+
url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
|
| 44 |
+
}
|
| 45 |
+
|
meta.yaml
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Combined metadata for all language-specific models
|
| 2 |
+
models:
|
| 3 |
+
bh:
|
| 4 |
+
espnet: '202412'
|
| 5 |
+
files:
|
| 6 |
+
asr_model_file: exp_small/exp_bh/asr_bh_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 7 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 8 |
+
timestamp: 1748120464.687666
|
| 9 |
+
torch: 2.3.0+cu121
|
| 10 |
+
yaml_files:
|
| 11 |
+
asr_train_config: exp_small/exp_bh/asr_bh_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 12 |
+
bn:
|
| 13 |
+
espnet: '202412'
|
| 14 |
+
files:
|
| 15 |
+
asr_model_file: exp_small/exp_bn/asr_bn_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 16 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 17 |
+
timestamp: 1748120485.2726
|
| 18 |
+
torch: 2.3.0+cu121
|
| 19 |
+
yaml_files:
|
| 20 |
+
asr_train_config: exp_small/exp_bn/asr_bn_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 21 |
+
ch:
|
| 22 |
+
espnet: '202412'
|
| 23 |
+
files:
|
| 24 |
+
asr_model_file: exp_small/exp_ch/asr_ch_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 25 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 26 |
+
timestamp: 1748120505.893883
|
| 27 |
+
torch: 2.3.0+cu121
|
| 28 |
+
yaml_files:
|
| 29 |
+
asr_train_config: exp_small/exp_ch/asr_ch_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 30 |
+
hi:
|
| 31 |
+
espnet: '202412'
|
| 32 |
+
files:
|
| 33 |
+
asr_model_file: exp_small/exp_hi/asr_hi_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 34 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 35 |
+
timestamp: 1748120526.738138
|
| 36 |
+
torch: 2.3.0+cu121
|
| 37 |
+
yaml_files:
|
| 38 |
+
asr_train_config: exp_small/exp_hi/asr_hi_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 39 |
+
kn:
|
| 40 |
+
espnet: '202412'
|
| 41 |
+
files:
|
| 42 |
+
asr_model_file: exp_small/exp_kn/asr_kn_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 43 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 44 |
+
timestamp: 1748124996.523497
|
| 45 |
+
torch: 2.3.0+cu121
|
| 46 |
+
yaml_files:
|
| 47 |
+
asr_train_config: exp_small/exp_kn/asr_kn_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 48 |
+
mg:
|
| 49 |
+
espnet: '202412'
|
| 50 |
+
files:
|
| 51 |
+
asr_model_file: exp_small/exp_mg/asr_mg_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 52 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 53 |
+
timestamp: 1748120568.092736
|
| 54 |
+
torch: 2.3.0+cu121
|
| 55 |
+
yaml_files:
|
| 56 |
+
asr_train_config: exp_small/exp_mg/asr_mg_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 57 |
+
mr:
|
| 58 |
+
espnet: '202412'
|
| 59 |
+
files:
|
| 60 |
+
asr_model_file: exp_small/exp_mr/asr_mr_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 61 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 62 |
+
timestamp: 1748120588.150078
|
| 63 |
+
torch: 2.3.0+cu121
|
| 64 |
+
yaml_files:
|
| 65 |
+
asr_train_config: exp_small/exp_mr/asr_mr_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 66 |
+
mt:
|
| 67 |
+
espnet: '202412'
|
| 68 |
+
files:
|
| 69 |
+
asr_model_file: exp_small/exp_mt/asr_mt_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_10best.pth
|
| 70 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 71 |
+
timestamp: 1748120608.234658
|
| 72 |
+
torch: 2.3.0+cu121
|
| 73 |
+
yaml_files:
|
| 74 |
+
asr_train_config: exp_small/exp_mt/asr_mt_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|
| 75 |
+
te:
|
| 76 |
+
espnet: '202412'
|
| 77 |
+
files:
|
| 78 |
+
asr_model_file: exp_small/exp_te/asr_te_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/valid.acc.ave_5best.pth
|
| 79 |
+
python: "3.8.10 (default, Mar 18 2025, 20:04:55) \n[GCC 9.4.0]"
|
| 80 |
+
timestamp: 1748120629.364632
|
| 81 |
+
torch: 2.3.0+cu121
|
| 82 |
+
yaml_files:
|
| 83 |
+
asr_train_config: exp_small/exp_te/asr_te_ebf_size256_mlp1024_lin1024_e8_mactrue_bs6M_gacc1/config.yaml
|