Commit
·
fb707a3
1
Parent(s):
8c887ec
Update model
Browse files- README.md +612 -0
- data_respin/data_asru25/nlsyms.txt +8 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/RESULTS.md +27 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/config.yaml +511 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/acc.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/backward_time.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/cer.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/cer_ctc.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/clip.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/forward_time.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/gpu_max_cached_mem_GB.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/grad_norm.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/iter_time.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss_att.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss_ctc.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss_scale.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/optim0_lr0.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/optim_step_time.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/train_time.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/wer.png +0 -0
- exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/valid.acc.ave_5best.pth +3 -0
- meta.yaml +8 -0
README.md
ADDED
@@ -0,0 +1,612 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- espnet
|
4 |
+
- audio
|
5 |
+
- automatic-speech-recognition
|
6 |
+
language: multilingual
|
7 |
+
datasets:
|
8 |
+
- respin_asru25
|
9 |
+
license: cc-by-4.0
|
10 |
+
---
|
11 |
+
|
12 |
+
## ESPnet2 ASR model
|
13 |
+
|
14 |
+
### `saurabhk0322/respin_asru25_track3`
|
15 |
+
|
16 |
+
This model was trained by wtc15 using respin_asru25 recipe in [espnet](https://github.com/espnet/espnet/).
|
17 |
+
|
18 |
+
### Demo: How to use in ESPnet2
|
19 |
+
|
20 |
+
Follow the [ESPnet installation instructions](https://espnet.github.io/espnet/installation.html)
|
21 |
+
if you haven't done that already.
|
22 |
+
|
23 |
+
```bash
|
24 |
+
cd espnet
|
25 |
+
|
26 |
+
pip install -e .
|
27 |
+
cd egs2/respin_asru25/asr1
|
28 |
+
./run.sh --skip_data_prep false --skip_train true --download_model saurabhk0322/respin_asru25_track3
|
29 |
+
```
|
30 |
+
|
31 |
+
<!-- Generated by scripts/utils/show_asr_result.sh -->
|
32 |
+
# RESULTS
|
33 |
+
## Environments
|
34 |
+
- date: `Wed Apr 23 00:04:53 IST 2025`
|
35 |
+
- python version: `3.10.12 (main, Feb 4 2025, 14:57:36) [GCC 11.4.0]`
|
36 |
+
- espnet version: `espnet 202503`
|
37 |
+
- pytorch version: `pytorch 2.3.0+cu121`
|
38 |
+
- Git hash: `0cc9d62673c1461efe37632aeab297a311bcd7f0`
|
39 |
+
- Commit date: `Sat Apr 12 17:53:44 2025 -0400`
|
40 |
+
|
41 |
+
## exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/decode_lid_asr_model_valid.acc.ave
|
42 |
+
### WER
|
43 |
+
|
44 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
45 |
+
|---|---|---|---|---|---|---|---|---|
|
46 |
+
|org/dev_lid|11507|111296|85.3|13.6|1.1|1.0|15.7|67.6|
|
47 |
+
|
48 |
+
### CER
|
49 |
+
|
50 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
51 |
+
|---|---|---|---|---|---|---|---|---|
|
52 |
+
|org/dev_lid|11507|615241|97.3|1.6|1.0|0.9|3.5|67.6|
|
53 |
+
|
54 |
+
### TER
|
55 |
+
|
56 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
57 |
+
|---|---|---|---|---|---|---|---|---|
|
58 |
+
|
59 |
+
## ASR config
|
60 |
+
|
61 |
+
<details><summary>expand</summary>
|
62 |
+
|
63 |
+
```
|
64 |
+
config: conf/tuning/train_asr_conformer_transformer_e8_linear1024_bs6M_gacc1.yaml
|
65 |
+
print_config: false
|
66 |
+
log_level: INFO
|
67 |
+
drop_last_iter: false
|
68 |
+
dry_run: false
|
69 |
+
iterator_type: sequence
|
70 |
+
valid_iterator_type: null
|
71 |
+
output_dir: exp/asru25/exp_large_lid_indic_char/asr_noaux_con_e8_lin1024_bs6M_gacc1_ctc03_lid
|
72 |
+
ngpu: 1
|
73 |
+
seed: 2022
|
74 |
+
num_workers: 8
|
75 |
+
num_att_plot: 3
|
76 |
+
dist_backend: nccl
|
77 |
+
dist_init_method: env://
|
78 |
+
dist_world_size: null
|
79 |
+
dist_rank: null
|
80 |
+
local_rank: 0
|
81 |
+
dist_master_addr: null
|
82 |
+
dist_master_port: null
|
83 |
+
dist_launcher: null
|
84 |
+
multiprocessing_distributed: false
|
85 |
+
unused_parameters: false
|
86 |
+
sharded_ddp: false
|
87 |
+
use_deepspeed: false
|
88 |
+
deepspeed_config: null
|
89 |
+
gradient_as_bucket_view: true
|
90 |
+
ddp_comm_hook: null
|
91 |
+
cudnn_enabled: true
|
92 |
+
cudnn_benchmark: false
|
93 |
+
cudnn_deterministic: true
|
94 |
+
use_tf32: false
|
95 |
+
collect_stats: false
|
96 |
+
write_collected_feats: false
|
97 |
+
max_epoch: 70
|
98 |
+
patience: 5
|
99 |
+
val_scheduler_criterion:
|
100 |
+
- valid
|
101 |
+
- loss
|
102 |
+
early_stopping_criterion:
|
103 |
+
- valid
|
104 |
+
- loss
|
105 |
+
- min
|
106 |
+
best_model_criterion:
|
107 |
+
- - valid
|
108 |
+
- acc
|
109 |
+
- max
|
110 |
+
keep_nbest_models: 5
|
111 |
+
nbest_averaging_interval: 0
|
112 |
+
grad_clip: 5.0
|
113 |
+
grad_clip_type: 2.0
|
114 |
+
grad_noise: false
|
115 |
+
accum_grad: 1
|
116 |
+
no_forward_run: false
|
117 |
+
resume: true
|
118 |
+
train_dtype: float32
|
119 |
+
use_amp: true
|
120 |
+
log_interval: null
|
121 |
+
use_matplotlib: true
|
122 |
+
use_tensorboard: true
|
123 |
+
create_graph_in_tensorboard: false
|
124 |
+
use_wandb: false
|
125 |
+
wandb_project: null
|
126 |
+
wandb_id: null
|
127 |
+
wandb_entity: null
|
128 |
+
wandb_name: null
|
129 |
+
wandb_model_log_interval: -1
|
130 |
+
detect_anomaly: false
|
131 |
+
use_adapter: false
|
132 |
+
adapter: lora
|
133 |
+
save_strategy: all
|
134 |
+
adapter_conf: {}
|
135 |
+
pretrain_path: null
|
136 |
+
init_param: []
|
137 |
+
ignore_init_mismatch: false
|
138 |
+
freeze_param: []
|
139 |
+
num_iters_per_epoch: null
|
140 |
+
batch_size: 20
|
141 |
+
valid_batch_size: null
|
142 |
+
batch_bins: 6000000
|
143 |
+
valid_batch_bins: null
|
144 |
+
category_sample_size: 10
|
145 |
+
train_shape_file:
|
146 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/train/speech_shape
|
147 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/train/text_shape.char
|
148 |
+
valid_shape_file:
|
149 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/valid/speech_shape
|
150 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/valid/text_shape.char
|
151 |
+
batch_type: numel
|
152 |
+
valid_batch_type: null
|
153 |
+
fold_length:
|
154 |
+
- 80000
|
155 |
+
- 150
|
156 |
+
sort_in_batch: descending
|
157 |
+
shuffle_within_batch: false
|
158 |
+
sort_batch: descending
|
159 |
+
multiple_iterator: false
|
160 |
+
chunk_length: 500
|
161 |
+
chunk_shift_ratio: 0.5
|
162 |
+
num_cache_chunks: 1024
|
163 |
+
chunk_excluded_key_prefixes: []
|
164 |
+
chunk_default_fs: null
|
165 |
+
chunk_max_abs_length: null
|
166 |
+
chunk_discard_short_samples: true
|
167 |
+
train_data_path_and_name_and_type:
|
168 |
+
- - dump/asru25_lid/raw/train_large_lid_sp/wav.scp
|
169 |
+
- speech
|
170 |
+
- sound
|
171 |
+
- - dump/asru25_lid/raw/train_large_lid_sp/text
|
172 |
+
- text
|
173 |
+
- text
|
174 |
+
valid_data_path_and_name_and_type:
|
175 |
+
- - dump/asru25_lid/raw/dev_lid/wav.scp
|
176 |
+
- speech
|
177 |
+
- sound
|
178 |
+
- - dump/asru25_lid/raw/dev_lid/text
|
179 |
+
- text
|
180 |
+
- text
|
181 |
+
multi_task_dataset: false
|
182 |
+
allow_variable_data_keys: false
|
183 |
+
max_cache_size: 0.0
|
184 |
+
max_cache_fd: 32
|
185 |
+
allow_multi_rates: false
|
186 |
+
valid_max_cache_size: null
|
187 |
+
exclude_weight_decay: false
|
188 |
+
exclude_weight_decay_conf: {}
|
189 |
+
optim: adam
|
190 |
+
optim_conf:
|
191 |
+
lr: 0.002
|
192 |
+
weight_decay: 1.0e-06
|
193 |
+
scheduler: warmuplr
|
194 |
+
scheduler_conf:
|
195 |
+
warmup_steps: 15000
|
196 |
+
token_list:
|
197 |
+
- <blank>
|
198 |
+
- <unk>
|
199 |
+
- <space>
|
200 |
+
- ा
|
201 |
+
- क
|
202 |
+
- े
|
203 |
+
- र
|
204 |
+
- ल
|
205 |
+
- न
|
206 |
+
- स
|
207 |
+
- ्
|
208 |
+
- त
|
209 |
+
- '['
|
210 |
+
- ']'
|
211 |
+
- म
|
212 |
+
- ी
|
213 |
+
- ि
|
214 |
+
- ್
|
215 |
+
- ह
|
216 |
+
- य
|
217 |
+
- ब
|
218 |
+
- प
|
219 |
+
- ो
|
220 |
+
- ్
|
221 |
+
- া
|
222 |
+
- व
|
223 |
+
- ज
|
224 |
+
- ं
|
225 |
+
- ు
|
226 |
+
- ��
|
227 |
+
- ి
|
228 |
+
- ಿ
|
229 |
+
- ে
|
230 |
+
- ಾ
|
231 |
+
- द
|
232 |
+
- র
|
233 |
+
- న
|
234 |
+
- ల
|
235 |
+
- ग
|
236 |
+
- ರ
|
237 |
+
- m
|
238 |
+
- ం
|
239 |
+
- క
|
240 |
+
- ক
|
241 |
+
- ర
|
242 |
+
- ು
|
243 |
+
- ನ
|
244 |
+
- ্
|
245 |
+
- ಕ
|
246 |
+
- ु
|
247 |
+
- च
|
248 |
+
- ट
|
249 |
+
- ತ
|
250 |
+
- ै
|
251 |
+
- ದ
|
252 |
+
- ি
|
253 |
+
- ख
|
254 |
+
- ೆ
|
255 |
+
- ಗ
|
256 |
+
- t
|
257 |
+
- ప
|
258 |
+
- త
|
259 |
+
- b
|
260 |
+
- h
|
261 |
+
- इ
|
262 |
+
- अ
|
263 |
+
- n
|
264 |
+
- आ
|
265 |
+
- ू
|
266 |
+
- ন
|
267 |
+
- ಸ
|
268 |
+
- ಲ
|
269 |
+
- వ
|
270 |
+
- ట
|
271 |
+
- స
|
272 |
+
- ವ
|
273 |
+
- ే
|
274 |
+
- छ
|
275 |
+
- ব
|
276 |
+
- ল
|
277 |
+
- ಯ
|
278 |
+
- भ
|
279 |
+
- श
|
280 |
+
- ಂ
|
281 |
+
- ಮ
|
282 |
+
- య
|
283 |
+
- য
|
284 |
+
- ध
|
285 |
+
- డ
|
286 |
+
- ద
|
287 |
+
- ए
|
288 |
+
- थ
|
289 |
+
- మ
|
290 |
+
- ಬ
|
291 |
+
- చ
|
292 |
+
- ण
|
293 |
+
- ड
|
294 |
+
- ম
|
295 |
+
- ई
|
296 |
+
- उ
|
297 |
+
- স
|
298 |
+
- ত
|
299 |
+
- ಳ
|
300 |
+
- ో
|
301 |
+
- ಡ
|
302 |
+
- फ
|
303 |
+
- g
|
304 |
+
- r
|
305 |
+
- e
|
306 |
+
- గ
|
307 |
+
- প
|
308 |
+
- ট
|
309 |
+
- য়
|
310 |
+
- c
|
311 |
+
- k
|
312 |
+
- ಟ
|
313 |
+
- ె
|
314 |
+
- .
|
315 |
+
- ಹ
|
316 |
+
- ಪ
|
317 |
+
- ೇ
|
318 |
+
- బ
|
319 |
+
- হ
|
320 |
+
- ু
|
321 |
+
- দ
|
322 |
+
- ष
|
323 |
+
- ো
|
324 |
+
- ీ
|
325 |
+
- জ
|
326 |
+
- ड़
|
327 |
+
- ಅ
|
328 |
+
- ೊ
|
329 |
+
- ই
|
330 |
+
- গ
|
331 |
+
- అ
|
332 |
+
- घ
|
333 |
+
- ಣ
|
334 |
+
- ठ
|
335 |
+
- ೋ
|
336 |
+
- চ
|
337 |
+
- ँ
|
338 |
+
- ొ
|
339 |
+
- ौ
|
340 |
+
- ছ
|
341 |
+
- ఎ
|
342 |
+
- ओ
|
343 |
+
- শ
|
344 |
+
- আ
|
345 |
+
- ూ
|
346 |
+
- జ
|
347 |
+
- ಜ
|
348 |
+
- থ
|
349 |
+
- ভ
|
350 |
+
- ಇ
|
351 |
+
- ೂ
|
352 |
+
- ಷ
|
353 |
+
- ಚ
|
354 |
+
- এ
|
355 |
+
- ষ
|
356 |
+
- ై
|
357 |
+
- ೀ
|
358 |
+
- ఉ
|
359 |
+
- ಎ
|
360 |
+
- ಆ
|
361 |
+
- ळ
|
362 |
+
- ধ
|
363 |
+
- ृ
|
364 |
+
- ী
|
365 |
+
- উ
|
366 |
+
- ফ
|
367 |
+
- খ
|
368 |
+
- ড
|
369 |
+
- ॉ
|
370 |
+
- ಶ
|
371 |
+
- অ
|
372 |
+
- ೈ
|
373 |
+
- ధ
|
374 |
+
- ং
|
375 |
+
- झ
|
376 |
+
- ఇ
|
377 |
+
- ఆ
|
378 |
+
- ష
|
379 |
+
- ढ
|
380 |
+
- ढ़
|
381 |
+
- భ
|
382 |
+
- శ
|
383 |
+
- ఏ
|
384 |
+
- ಧ
|
385 |
+
- ও
|
386 |
+
- ಒ
|
387 |
+
- ಭ
|
388 |
+
- ళ
|
389 |
+
- ಉ
|
390 |
+
- ॅ
|
391 |
+
- ಫ
|
392 |
+
- ऊ
|
393 |
+
- ఫ
|
394 |
+
- ಥ
|
395 |
+
- ऑ
|
396 |
+
- ణ
|
397 |
+
- ড়
|
398 |
+
- ণ
|
399 |
+
- ঙ
|
400 |
+
- ऋ
|
401 |
+
- ಖ
|
402 |
+
- ऽ
|
403 |
+
- హ
|
404 |
+
- థ
|
405 |
+
- औ
|
406 |
+
- ೃ
|
407 |
+
- ঁ
|
408 |
+
- ೌ
|
409 |
+
- ఒ
|
410 |
+
- ఖ
|
411 |
+
- ৃ
|
412 |
+
- ఈ
|
413 |
+
- ಏ
|
414 |
+
- ঠ
|
415 |
+
- ౌ
|
416 |
+
- ಐ
|
417 |
+
- ৈ
|
418 |
+
- ऐ
|
419 |
+
- ऱ
|
420 |
+
- ఐ
|
421 |
+
- ূ
|
422 |
+
- ञ
|
423 |
+
- ৎ
|
424 |
+
- ృ
|
425 |
+
- ज़
|
426 |
+
- ঞ
|
427 |
+
- ಈ
|
428 |
+
- ঘ
|
429 |
+
- ঋ
|
430 |
+
- ঝ
|
431 |
+
- फ़
|
432 |
+
- ৌ
|
433 |
+
- ಠ
|
434 |
+
- ः
|
435 |
+
- ఓ
|
436 |
+
- ಘ
|
437 |
+
- ಛ
|
438 |
+
- ಓ
|
439 |
+
- ఊ
|
440 |
+
- ఋ
|
441 |
+
- ಔ
|
442 |
+
- ఛ
|
443 |
+
- ಞ
|
444 |
+
- ॲ
|
445 |
+
- ಊ
|
446 |
+
- ఘ
|
447 |
+
- ঢ
|
448 |
+
- ख़
|
449 |
+
- ়
|
450 |
+
- ऍ
|
451 |
+
- ಋ
|
452 |
+
- क़
|
453 |
+
- ఠ
|
454 |
+
- ঢ়
|
455 |
+
- ঃ
|
456 |
+
- ़
|
457 |
+
- ೕ
|
458 |
+
- ఔ
|
459 |
+
- ಢ
|
460 |
+
- ঊ
|
461 |
+
- ఱ
|
462 |
+
- ಃ
|
463 |
+
- ಝ
|
464 |
+
- ङ
|
465 |
+
- ఢ
|
466 |
+
- ग़
|
467 |
+
- ఞ
|
468 |
+
- ঐ
|
469 |
+
- ঔ
|
470 |
+
- ॠ
|
471 |
+
- ':'
|
472 |
+
- ೯
|
473 |
+
- ೖ
|
474 |
+
- ঈ
|
475 |
+
- ః
|
476 |
+
- ౖ
|
477 |
+
- ৠ
|
478 |
+
- ౦
|
479 |
+
- <sos/eos>
|
480 |
+
init: null
|
481 |
+
input_size: null
|
482 |
+
ctc_conf:
|
483 |
+
dropout_rate: 0.0
|
484 |
+
ctc_type: builtin
|
485 |
+
reduce: true
|
486 |
+
ignore_nan_grad: null
|
487 |
+
zero_infinity: true
|
488 |
+
brctc_risk_strategy: exp
|
489 |
+
brctc_group_strategy: end
|
490 |
+
brctc_risk_factor: 0.0
|
491 |
+
joint_net_conf: null
|
492 |
+
use_preprocessor: true
|
493 |
+
use_lang_prompt: false
|
494 |
+
use_nlp_prompt: false
|
495 |
+
token_type: char
|
496 |
+
bpemodel: null
|
497 |
+
non_linguistic_symbols: data/nlsyms.txt
|
498 |
+
cleaner: null
|
499 |
+
g2p: null
|
500 |
+
speech_volume_normalize: null
|
501 |
+
rir_scp: null
|
502 |
+
rir_apply_prob: 1.0
|
503 |
+
noise_scp: null
|
504 |
+
noise_apply_prob: 1.0
|
505 |
+
noise_db_range: '13_15'
|
506 |
+
short_noise_thres: 0.5
|
507 |
+
aux_ctc_tasks: []
|
508 |
+
frontend: default
|
509 |
+
frontend_conf:
|
510 |
+
n_fft: 512
|
511 |
+
win_length: 400
|
512 |
+
hop_length: 160
|
513 |
+
fs: 16k
|
514 |
+
specaug: specaug
|
515 |
+
specaug_conf:
|
516 |
+
apply_time_warp: true
|
517 |
+
time_warp_window: 5
|
518 |
+
time_warp_mode: bicubic
|
519 |
+
apply_freq_mask: true
|
520 |
+
freq_mask_width_range:
|
521 |
+
- 0
|
522 |
+
- 27
|
523 |
+
num_freq_mask: 2
|
524 |
+
apply_time_mask: true
|
525 |
+
time_mask_width_ratio_range:
|
526 |
+
- 0.0
|
527 |
+
- 0.05
|
528 |
+
num_time_mask: 5
|
529 |
+
normalize: utterance_mvn
|
530 |
+
normalize_conf: {}
|
531 |
+
model: espnet
|
532 |
+
model_conf:
|
533 |
+
ctc_weight: 0.3
|
534 |
+
lsm_weight: 0.1
|
535 |
+
length_normalized_loss: false
|
536 |
+
preencoder: null
|
537 |
+
preencoder_conf: {}
|
538 |
+
encoder: conformer
|
539 |
+
encoder_conf:
|
540 |
+
output_size: 256
|
541 |
+
attention_heads: 4
|
542 |
+
linear_units: 1024
|
543 |
+
num_blocks: 8
|
544 |
+
dropout_rate: 0.1
|
545 |
+
positional_dropout_rate: 0.1
|
546 |
+
attention_dropout_rate: 0.1
|
547 |
+
input_layer: conv2d2
|
548 |
+
normalize_before: true
|
549 |
+
macaron_style: true
|
550 |
+
rel_pos_type: latest
|
551 |
+
pos_enc_layer_type: rel_pos
|
552 |
+
selfattention_layer_type: rel_selfattn
|
553 |
+
activation_type: swish
|
554 |
+
use_cnn_module: true
|
555 |
+
cnn_module_kernel: 31
|
556 |
+
postencoder: null
|
557 |
+
postencoder_conf: {}
|
558 |
+
decoder: transformer
|
559 |
+
decoder_conf:
|
560 |
+
attention_heads: 4
|
561 |
+
linear_units: 2048
|
562 |
+
num_blocks: 6
|
563 |
+
dropout_rate: 0.1
|
564 |
+
positional_dropout_rate: 0.1
|
565 |
+
self_attention_dropout_rate: 0.1
|
566 |
+
src_attention_dropout_rate: 0.1
|
567 |
+
layer_drop_rate: 0.0
|
568 |
+
preprocessor: default
|
569 |
+
preprocessor_conf: {}
|
570 |
+
required:
|
571 |
+
- output_dir
|
572 |
+
- token_list
|
573 |
+
version: '202503'
|
574 |
+
distributed: false
|
575 |
+
```
|
576 |
+
|
577 |
+
</details>
|
578 |
+
|
579 |
+
|
580 |
+
|
581 |
+
### Citing ESPnet
|
582 |
+
|
583 |
+
```BibTex
|
584 |
+
@inproceedings{watanabe2018espnet,
|
585 |
+
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},
|
586 |
+
title={{ESPnet}: End-to-End Speech Processing Toolkit},
|
587 |
+
year={2018},
|
588 |
+
booktitle={Proceedings of Interspeech},
|
589 |
+
pages={2207--2211},
|
590 |
+
doi={10.21437/Interspeech.2018-1456},
|
591 |
+
url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
|
592 |
+
}
|
593 |
+
|
594 |
+
|
595 |
+
|
596 |
+
|
597 |
+
|
598 |
+
|
599 |
+
```
|
600 |
+
|
601 |
+
or arXiv:
|
602 |
+
|
603 |
+
```bibtex
|
604 |
+
@misc{watanabe2018espnet,
|
605 |
+
title={ESPnet: End-to-End Speech Processing Toolkit},
|
606 |
+
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},
|
607 |
+
year={2018},
|
608 |
+
eprint={1804.00015},
|
609 |
+
archivePrefix={arXiv},
|
610 |
+
primaryClass={cs.CL}
|
611 |
+
}
|
612 |
+
```
|
data_respin/data_asru25/nlsyms.txt
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
[bh]
|
2 |
+
[bn]
|
3 |
+
[ch]
|
4 |
+
[kn]
|
5 |
+
[mg]
|
6 |
+
[mr]
|
7 |
+
[mt]
|
8 |
+
[te]
|
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/RESULTS.md
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<!-- Generated by scripts/utils/show_asr_result.sh -->
|
2 |
+
# RESULTS
|
3 |
+
## Environments
|
4 |
+
- date: `Wed Apr 23 00:04:53 IST 2025`
|
5 |
+
- python version: `3.10.12 (main, Feb 4 2025, 14:57:36) [GCC 11.4.0]`
|
6 |
+
- espnet version: `espnet 202503`
|
7 |
+
- pytorch version: `pytorch 2.3.0+cu121`
|
8 |
+
- Git hash: `0cc9d62673c1461efe37632aeab297a311bcd7f0`
|
9 |
+
- Commit date: `Sat Apr 12 17:53:44 2025 -0400`
|
10 |
+
|
11 |
+
## exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/decode_lid_asr_model_valid.acc.ave
|
12 |
+
### WER
|
13 |
+
|
14 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
15 |
+
|---|---|---|---|---|---|---|---|---|
|
16 |
+
|org/dev_lid|11507|111296|85.3|13.6|1.1|1.0|15.7|67.6|
|
17 |
+
|
18 |
+
### CER
|
19 |
+
|
20 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
21 |
+
|---|---|---|---|---|---|---|---|---|
|
22 |
+
|org/dev_lid|11507|615241|97.3|1.6|1.0|0.9|3.5|67.6|
|
23 |
+
|
24 |
+
### TER
|
25 |
+
|
26 |
+
|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
|
27 |
+
|---|---|---|---|---|---|---|---|---|
|
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/config.yaml
ADDED
@@ -0,0 +1,511 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf/tuning/train_asr_conformer_transformer_e8_linear1024_bs6M_gacc1.yaml
|
2 |
+
print_config: false
|
3 |
+
log_level: INFO
|
4 |
+
drop_last_iter: false
|
5 |
+
dry_run: false
|
6 |
+
iterator_type: sequence
|
7 |
+
valid_iterator_type: null
|
8 |
+
output_dir: exp/asru25/exp_large_lid_indic_char/asr_noaux_con_e8_lin1024_bs6M_gacc1_ctc03_lid
|
9 |
+
ngpu: 1
|
10 |
+
seed: 2022
|
11 |
+
num_workers: 8
|
12 |
+
num_att_plot: 3
|
13 |
+
dist_backend: nccl
|
14 |
+
dist_init_method: env://
|
15 |
+
dist_world_size: null
|
16 |
+
dist_rank: null
|
17 |
+
local_rank: 0
|
18 |
+
dist_master_addr: null
|
19 |
+
dist_master_port: null
|
20 |
+
dist_launcher: null
|
21 |
+
multiprocessing_distributed: false
|
22 |
+
unused_parameters: false
|
23 |
+
sharded_ddp: false
|
24 |
+
use_deepspeed: false
|
25 |
+
deepspeed_config: null
|
26 |
+
gradient_as_bucket_view: true
|
27 |
+
ddp_comm_hook: null
|
28 |
+
cudnn_enabled: true
|
29 |
+
cudnn_benchmark: false
|
30 |
+
cudnn_deterministic: true
|
31 |
+
use_tf32: false
|
32 |
+
collect_stats: false
|
33 |
+
write_collected_feats: false
|
34 |
+
max_epoch: 70
|
35 |
+
patience: 5
|
36 |
+
val_scheduler_criterion:
|
37 |
+
- valid
|
38 |
+
- loss
|
39 |
+
early_stopping_criterion:
|
40 |
+
- valid
|
41 |
+
- loss
|
42 |
+
- min
|
43 |
+
best_model_criterion:
|
44 |
+
- - valid
|
45 |
+
- acc
|
46 |
+
- max
|
47 |
+
keep_nbest_models: 5
|
48 |
+
nbest_averaging_interval: 0
|
49 |
+
grad_clip: 5.0
|
50 |
+
grad_clip_type: 2.0
|
51 |
+
grad_noise: false
|
52 |
+
accum_grad: 1
|
53 |
+
no_forward_run: false
|
54 |
+
resume: true
|
55 |
+
train_dtype: float32
|
56 |
+
use_amp: true
|
57 |
+
log_interval: null
|
58 |
+
use_matplotlib: true
|
59 |
+
use_tensorboard: true
|
60 |
+
create_graph_in_tensorboard: false
|
61 |
+
use_wandb: false
|
62 |
+
wandb_project: null
|
63 |
+
wandb_id: null
|
64 |
+
wandb_entity: null
|
65 |
+
wandb_name: null
|
66 |
+
wandb_model_log_interval: -1
|
67 |
+
detect_anomaly: false
|
68 |
+
use_adapter: false
|
69 |
+
adapter: lora
|
70 |
+
save_strategy: all
|
71 |
+
adapter_conf: {}
|
72 |
+
pretrain_path: null
|
73 |
+
init_param: []
|
74 |
+
ignore_init_mismatch: false
|
75 |
+
freeze_param: []
|
76 |
+
num_iters_per_epoch: null
|
77 |
+
batch_size: 20
|
78 |
+
valid_batch_size: null
|
79 |
+
batch_bins: 6000000
|
80 |
+
valid_batch_bins: null
|
81 |
+
category_sample_size: 10
|
82 |
+
train_shape_file:
|
83 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/train/speech_shape
|
84 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/train/text_shape.char
|
85 |
+
valid_shape_file:
|
86 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/valid/speech_shape
|
87 |
+
- exp/asru25/exp_large_lid_indic_char/asr_stats_raw_large_lid_char_sp/valid/text_shape.char
|
88 |
+
batch_type: numel
|
89 |
+
valid_batch_type: null
|
90 |
+
fold_length:
|
91 |
+
- 80000
|
92 |
+
- 150
|
93 |
+
sort_in_batch: descending
|
94 |
+
shuffle_within_batch: false
|
95 |
+
sort_batch: descending
|
96 |
+
multiple_iterator: false
|
97 |
+
chunk_length: 500
|
98 |
+
chunk_shift_ratio: 0.5
|
99 |
+
num_cache_chunks: 1024
|
100 |
+
chunk_excluded_key_prefixes: []
|
101 |
+
chunk_default_fs: null
|
102 |
+
chunk_max_abs_length: null
|
103 |
+
chunk_discard_short_samples: true
|
104 |
+
train_data_path_and_name_and_type:
|
105 |
+
- - dump/asru25_lid/raw/train_large_lid_sp/wav.scp
|
106 |
+
- speech
|
107 |
+
- sound
|
108 |
+
- - dump/asru25_lid/raw/train_large_lid_sp/text
|
109 |
+
- text
|
110 |
+
- text
|
111 |
+
valid_data_path_and_name_and_type:
|
112 |
+
- - dump/asru25_lid/raw/dev_lid/wav.scp
|
113 |
+
- speech
|
114 |
+
- sound
|
115 |
+
- - dump/asru25_lid/raw/dev_lid/text
|
116 |
+
- text
|
117 |
+
- text
|
118 |
+
multi_task_dataset: false
|
119 |
+
allow_variable_data_keys: false
|
120 |
+
max_cache_size: 0.0
|
121 |
+
max_cache_fd: 32
|
122 |
+
allow_multi_rates: false
|
123 |
+
valid_max_cache_size: null
|
124 |
+
exclude_weight_decay: false
|
125 |
+
exclude_weight_decay_conf: {}
|
126 |
+
optim: adam
|
127 |
+
optim_conf:
|
128 |
+
lr: 0.002
|
129 |
+
weight_decay: 1.0e-06
|
130 |
+
scheduler: warmuplr
|
131 |
+
scheduler_conf:
|
132 |
+
warmup_steps: 15000
|
133 |
+
token_list:
|
134 |
+
- <blank>
|
135 |
+
- <unk>
|
136 |
+
- <space>
|
137 |
+
- ा
|
138 |
+
- क
|
139 |
+
- े
|
140 |
+
- र
|
141 |
+
- ल
|
142 |
+
- न
|
143 |
+
- स
|
144 |
+
- ्
|
145 |
+
- त
|
146 |
+
- '['
|
147 |
+
- ']'
|
148 |
+
- म
|
149 |
+
- ी
|
150 |
+
- ि
|
151 |
+
- ್
|
152 |
+
- ह
|
153 |
+
- य
|
154 |
+
- ब
|
155 |
+
- प
|
156 |
+
- ो
|
157 |
+
- ్
|
158 |
+
- া
|
159 |
+
- व
|
160 |
+
- ज
|
161 |
+
- ं
|
162 |
+
- ు
|
163 |
+
- ా
|
164 |
+
- ి
|
165 |
+
- ಿ
|
166 |
+
- ে
|
167 |
+
- ಾ
|
168 |
+
- द
|
169 |
+
- র
|
170 |
+
- న
|
171 |
+
- ల
|
172 |
+
- ग
|
173 |
+
- ರ
|
174 |
+
- m
|
175 |
+
- ం
|
176 |
+
- క
|
177 |
+
- ক
|
178 |
+
- ర
|
179 |
+
- ು
|
180 |
+
- ನ
|
181 |
+
- ্
|
182 |
+
- ಕ
|
183 |
+
- ु
|
184 |
+
- च
|
185 |
+
- ट
|
186 |
+
- ತ
|
187 |
+
- ै
|
188 |
+
- ದ
|
189 |
+
- ি
|
190 |
+
- ख
|
191 |
+
- ೆ
|
192 |
+
- ಗ
|
193 |
+
- t
|
194 |
+
- ప
|
195 |
+
- త
|
196 |
+
- b
|
197 |
+
- h
|
198 |
+
- इ
|
199 |
+
- अ
|
200 |
+
- n
|
201 |
+
- आ
|
202 |
+
- ू
|
203 |
+
- ন
|
204 |
+
- ಸ
|
205 |
+
- ಲ
|
206 |
+
- వ
|
207 |
+
- ట
|
208 |
+
- స
|
209 |
+
- ವ
|
210 |
+
- ే
|
211 |
+
- छ
|
212 |
+
- ব
|
213 |
+
- ল
|
214 |
+
- ಯ
|
215 |
+
- भ
|
216 |
+
- श
|
217 |
+
- ಂ
|
218 |
+
- ಮ
|
219 |
+
- య
|
220 |
+
- য
|
221 |
+
- ध
|
222 |
+
- డ
|
223 |
+
- ద
|
224 |
+
- ए
|
225 |
+
- थ
|
226 |
+
- మ
|
227 |
+
- ಬ
|
228 |
+
- చ
|
229 |
+
- ण
|
230 |
+
- ड
|
231 |
+
- ম
|
232 |
+
- ई
|
233 |
+
- उ
|
234 |
+
- স
|
235 |
+
- ত
|
236 |
+
- ಳ
|
237 |
+
- ో
|
238 |
+
- ಡ
|
239 |
+
- फ
|
240 |
+
- g
|
241 |
+
- r
|
242 |
+
- e
|
243 |
+
- గ
|
244 |
+
- প
|
245 |
+
- ট
|
246 |
+
- য়
|
247 |
+
- c
|
248 |
+
- k
|
249 |
+
- ಟ
|
250 |
+
- ె
|
251 |
+
- .
|
252 |
+
- ಹ
|
253 |
+
- ಪ
|
254 |
+
- ೇ
|
255 |
+
- బ
|
256 |
+
- হ
|
257 |
+
- ু
|
258 |
+
- দ
|
259 |
+
- ष
|
260 |
+
- ো
|
261 |
+
- ీ
|
262 |
+
- জ
|
263 |
+
- ड़
|
264 |
+
- ಅ
|
265 |
+
- ೊ
|
266 |
+
- ই
|
267 |
+
- গ
|
268 |
+
- అ
|
269 |
+
- घ
|
270 |
+
- ಣ
|
271 |
+
- ठ
|
272 |
+
- ೋ
|
273 |
+
- চ
|
274 |
+
- ँ
|
275 |
+
- ొ
|
276 |
+
- ौ
|
277 |
+
- ছ
|
278 |
+
- ఎ
|
279 |
+
- ओ
|
280 |
+
- শ
|
281 |
+
- আ
|
282 |
+
- ూ
|
283 |
+
- జ
|
284 |
+
- ಜ
|
285 |
+
- থ
|
286 |
+
- ভ
|
287 |
+
- ಇ
|
288 |
+
- ೂ
|
289 |
+
- ಷ
|
290 |
+
- ಚ
|
291 |
+
- এ
|
292 |
+
- ষ
|
293 |
+
- ై
|
294 |
+
- ೀ
|
295 |
+
- ఉ
|
296 |
+
- ಎ
|
297 |
+
- ಆ
|
298 |
+
- ळ
|
299 |
+
- ধ
|
300 |
+
- ृ
|
301 |
+
- ী
|
302 |
+
- উ
|
303 |
+
- ফ
|
304 |
+
- খ
|
305 |
+
- ড
|
306 |
+
- ॉ
|
307 |
+
- ಶ
|
308 |
+
- অ
|
309 |
+
- ೈ
|
310 |
+
- ధ
|
311 |
+
- ং
|
312 |
+
- झ
|
313 |
+
- ఇ
|
314 |
+
- ఆ
|
315 |
+
- ష
|
316 |
+
- ढ
|
317 |
+
- ढ़
|
318 |
+
- భ
|
319 |
+
- శ
|
320 |
+
- ఏ
|
321 |
+
- ಧ
|
322 |
+
- ও
|
323 |
+
- ಒ
|
324 |
+
- ಭ
|
325 |
+
- ళ
|
326 |
+
- ಉ
|
327 |
+
- ॅ
|
328 |
+
- ಫ
|
329 |
+
- ऊ
|
330 |
+
- ఫ
|
331 |
+
- ಥ
|
332 |
+
- ऑ
|
333 |
+
- ణ
|
334 |
+
- ড়
|
335 |
+
- ণ
|
336 |
+
- ঙ
|
337 |
+
- ऋ
|
338 |
+
- ಖ
|
339 |
+
- ऽ
|
340 |
+
- హ
|
341 |
+
- థ
|
342 |
+
- औ
|
343 |
+
- ೃ
|
344 |
+
- ঁ
|
345 |
+
- ೌ
|
346 |
+
- ఒ
|
347 |
+
- ఖ
|
348 |
+
- ৃ
|
349 |
+
- ఈ
|
350 |
+
- ಏ
|
351 |
+
- ঠ
|
352 |
+
- ౌ
|
353 |
+
- ಐ
|
354 |
+
- ৈ
|
355 |
+
- ऐ
|
356 |
+
- ऱ
|
357 |
+
- ఐ
|
358 |
+
- ূ
|
359 |
+
- ञ
|
360 |
+
- ৎ
|
361 |
+
- ృ
|
362 |
+
- ज़
|
363 |
+
- ঞ
|
364 |
+
- ಈ
|
365 |
+
- ঘ
|
366 |
+
- ঋ
|
367 |
+
- ঝ
|
368 |
+
- फ़
|
369 |
+
- ৌ
|
370 |
+
- ಠ
|
371 |
+
- ः
|
372 |
+
- ఓ
|
373 |
+
- ಘ
|
374 |
+
- ಛ
|
375 |
+
- ಓ
|
376 |
+
- ఊ
|
377 |
+
- ఋ
|
378 |
+
- ಔ
|
379 |
+
- ఛ
|
380 |
+
- ಞ
|
381 |
+
- ॲ
|
382 |
+
- ಊ
|
383 |
+
- ఘ
|
384 |
+
- ঢ
|
385 |
+
- ख़
|
386 |
+
- ়
|
387 |
+
- ऍ
|
388 |
+
- ಋ
|
389 |
+
- क़
|
390 |
+
- ఠ
|
391 |
+
- ঢ়
|
392 |
+
- ঃ
|
393 |
+
- ़
|
394 |
+
- ೕ
|
395 |
+
- ఔ
|
396 |
+
- ಢ
|
397 |
+
- ঊ
|
398 |
+
- ఱ
|
399 |
+
- ಃ
|
400 |
+
- ಝ
|
401 |
+
- ङ
|
402 |
+
- ఢ
|
403 |
+
- ग़
|
404 |
+
- ఞ
|
405 |
+
- ঐ
|
406 |
+
- ঔ
|
407 |
+
- ॠ
|
408 |
+
- ':'
|
409 |
+
- ೯
|
410 |
+
- ೖ
|
411 |
+
- ঈ
|
412 |
+
- ః
|
413 |
+
- ౖ
|
414 |
+
- ৠ
|
415 |
+
- ౦
|
416 |
+
- <sos/eos>
|
417 |
+
init: null
|
418 |
+
input_size: null
|
419 |
+
ctc_conf:
|
420 |
+
dropout_rate: 0.0
|
421 |
+
ctc_type: builtin
|
422 |
+
reduce: true
|
423 |
+
ignore_nan_grad: null
|
424 |
+
zero_infinity: true
|
425 |
+
brctc_risk_strategy: exp
|
426 |
+
brctc_group_strategy: end
|
427 |
+
brctc_risk_factor: 0.0
|
428 |
+
joint_net_conf: null
|
429 |
+
use_preprocessor: true
|
430 |
+
use_lang_prompt: false
|
431 |
+
use_nlp_prompt: false
|
432 |
+
token_type: char
|
433 |
+
bpemodel: null
|
434 |
+
non_linguistic_symbols: data/nlsyms.txt
|
435 |
+
cleaner: null
|
436 |
+
g2p: null
|
437 |
+
speech_volume_normalize: null
|
438 |
+
rir_scp: null
|
439 |
+
rir_apply_prob: 1.0
|
440 |
+
noise_scp: null
|
441 |
+
noise_apply_prob: 1.0
|
442 |
+
noise_db_range: '13_15'
|
443 |
+
short_noise_thres: 0.5
|
444 |
+
aux_ctc_tasks: []
|
445 |
+
frontend: default
|
446 |
+
frontend_conf:
|
447 |
+
n_fft: 512
|
448 |
+
win_length: 400
|
449 |
+
hop_length: 160
|
450 |
+
fs: 16k
|
451 |
+
specaug: specaug
|
452 |
+
specaug_conf:
|
453 |
+
apply_time_warp: true
|
454 |
+
time_warp_window: 5
|
455 |
+
time_warp_mode: bicubic
|
456 |
+
apply_freq_mask: true
|
457 |
+
freq_mask_width_range:
|
458 |
+
- 0
|
459 |
+
- 27
|
460 |
+
num_freq_mask: 2
|
461 |
+
apply_time_mask: true
|
462 |
+
time_mask_width_ratio_range:
|
463 |
+
- 0.0
|
464 |
+
- 0.05
|
465 |
+
num_time_mask: 5
|
466 |
+
normalize: utterance_mvn
|
467 |
+
normalize_conf: {}
|
468 |
+
model: espnet
|
469 |
+
model_conf:
|
470 |
+
ctc_weight: 0.3
|
471 |
+
lsm_weight: 0.1
|
472 |
+
length_normalized_loss: false
|
473 |
+
preencoder: null
|
474 |
+
preencoder_conf: {}
|
475 |
+
encoder: conformer
|
476 |
+
encoder_conf:
|
477 |
+
output_size: 256
|
478 |
+
attention_heads: 4
|
479 |
+
linear_units: 1024
|
480 |
+
num_blocks: 8
|
481 |
+
dropout_rate: 0.1
|
482 |
+
positional_dropout_rate: 0.1
|
483 |
+
attention_dropout_rate: 0.1
|
484 |
+
input_layer: conv2d2
|
485 |
+
normalize_before: true
|
486 |
+
macaron_style: true
|
487 |
+
rel_pos_type: latest
|
488 |
+
pos_enc_layer_type: rel_pos
|
489 |
+
selfattention_layer_type: rel_selfattn
|
490 |
+
activation_type: swish
|
491 |
+
use_cnn_module: true
|
492 |
+
cnn_module_kernel: 31
|
493 |
+
postencoder: null
|
494 |
+
postencoder_conf: {}
|
495 |
+
decoder: transformer
|
496 |
+
decoder_conf:
|
497 |
+
attention_heads: 4
|
498 |
+
linear_units: 2048
|
499 |
+
num_blocks: 6
|
500 |
+
dropout_rate: 0.1
|
501 |
+
positional_dropout_rate: 0.1
|
502 |
+
self_attention_dropout_rate: 0.1
|
503 |
+
src_attention_dropout_rate: 0.1
|
504 |
+
layer_drop_rate: 0.0
|
505 |
+
preprocessor: default
|
506 |
+
preprocessor_conf: {}
|
507 |
+
required:
|
508 |
+
- output_dir
|
509 |
+
- token_list
|
510 |
+
version: '202503'
|
511 |
+
distributed: false
|
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/acc.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/backward_time.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/cer.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/cer_ctc.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/clip.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/forward_time.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/gpu_max_cached_mem_GB.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/grad_norm.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/iter_time.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss_att.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss_ctc.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/loss_scale.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/optim0_lr0.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/optim_step_time.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/train_time.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/images/wer.png
ADDED
![]() |
exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/valid.acc.ave_5best.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e95fdc2cca38ece938653d8033d2da3729981d780fd931ad56080593d63c5838
|
3 |
+
size 101971162
|
meta.yaml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
espnet: '202503'
|
2 |
+
files:
|
3 |
+
asr_model_file: exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/valid.acc.ave_5best.pth
|
4 |
+
python: 3.10.12 (main, Feb 4 2025, 14:57:36) [GCC 11.4.0]
|
5 |
+
timestamp: 1745346895.184633
|
6 |
+
torch: 2.3.0+cu121
|
7 |
+
yaml_files:
|
8 |
+
asr_train_config: exp/asru25/exp_large/asr_multilingual_lid_con_e8_lin1024_bs6M_gacc1_ctc03/config.yaml
|