Spaces:
Runtime error
Runtime error
Commit
·
97ff6ba
1
Parent(s):
82c3ef1
add wenet models
Browse files
model.py
CHANGED
|
@@ -327,6 +327,46 @@ def _get_alimeeting_pre_trained_model(
|
|
| 327 |
return recognizer
|
| 328 |
|
| 329 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 330 |
@lru_cache(maxsize=10)
|
| 331 |
def _get_aidatatang_200zh_pretrained_mode(
|
| 332 |
repo_id: str,
|
|
@@ -448,6 +488,7 @@ chinese_models = {
|
|
| 448 |
"yuekai/icefall-asr-aishell2-pruned-transducer-stateless5-B-2022-07-12": _get_aishell2_pretrained_model, # noqa
|
| 449 |
"luomingshuang/icefall_asr_aidatatang-200zh_pruned_transducer_stateless2": _get_aidatatang_200zh_pretrained_mode, # noqa
|
| 450 |
"luomingshuang/icefall_asr_alimeeting_pruned_transducer_stateless2": _get_alimeeting_pre_trained_model, # noqa
|
|
|
|
| 451 |
}
|
| 452 |
|
| 453 |
english_models = {
|
|
@@ -456,6 +497,7 @@ english_models = {
|
|
| 456 |
"csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless8-2022-11-14": _get_librispeech_pre_trained_model, # noqa
|
| 457 |
"csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless7-2022-11-11": _get_librispeech_pre_trained_model, # noqa
|
| 458 |
"csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13": _get_librispeech_pre_trained_model, # noqa
|
|
|
|
| 459 |
}
|
| 460 |
|
| 461 |
chinese_english_mixed_models = {
|
|
|
|
| 327 |
return recognizer
|
| 328 |
|
| 329 |
|
| 330 |
+
@lru_cache(maxsize=10)
|
| 331 |
+
def _get_wenet_model(
|
| 332 |
+
repo_id: str,
|
| 333 |
+
decoding_method: str,
|
| 334 |
+
num_active_paths: int,
|
| 335 |
+
):
|
| 336 |
+
assert repo_id in [
|
| 337 |
+
"csukuangfj/wenet-chinese-model",
|
| 338 |
+
], repo_id
|
| 339 |
+
|
| 340 |
+
nn_model = _get_nn_model_filename(
|
| 341 |
+
repo_id=repo_id,
|
| 342 |
+
filename="final.zip",
|
| 343 |
+
subfolder="./",
|
| 344 |
+
)
|
| 345 |
+
tokens = _get_token_filename(
|
| 346 |
+
repo_id=repo_id,
|
| 347 |
+
filename="units.txt",
|
| 348 |
+
subfolder="./",
|
| 349 |
+
)
|
| 350 |
+
|
| 351 |
+
feat_config = sherpa.FeatureConfig(normalize_samples=False)
|
| 352 |
+
feat_config.fbank_opts.frame_opts.samp_freq = sample_rate
|
| 353 |
+
feat_config.fbank_opts.mel_opts.num_bins = 80
|
| 354 |
+
feat_config.fbank_opts.frame_opts.dither = 0
|
| 355 |
+
|
| 356 |
+
config = sherpa.OfflineRecognizerConfig(
|
| 357 |
+
nn_model=nn_model,
|
| 358 |
+
tokens=tokens,
|
| 359 |
+
use_gpu=False,
|
| 360 |
+
feat_config=feat_config,
|
| 361 |
+
decoding_method=decoding_method,
|
| 362 |
+
num_active_paths=num_active_paths,
|
| 363 |
+
)
|
| 364 |
+
|
| 365 |
+
recognizer = sherpa.OfflineRecognizer(config)
|
| 366 |
+
|
| 367 |
+
return recognizer
|
| 368 |
+
|
| 369 |
+
|
| 370 |
@lru_cache(maxsize=10)
|
| 371 |
def _get_aidatatang_200zh_pretrained_mode(
|
| 372 |
repo_id: str,
|
|
|
|
| 488 |
"yuekai/icefall-asr-aishell2-pruned-transducer-stateless5-B-2022-07-12": _get_aishell2_pretrained_model, # noqa
|
| 489 |
"luomingshuang/icefall_asr_aidatatang-200zh_pruned_transducer_stateless2": _get_aidatatang_200zh_pretrained_mode, # noqa
|
| 490 |
"luomingshuang/icefall_asr_alimeeting_pruned_transducer_stateless2": _get_alimeeting_pre_trained_model, # noqa
|
| 491 |
+
"csukuangfj/wenet-chinese-model": _get_wenet_model,
|
| 492 |
}
|
| 493 |
|
| 494 |
english_models = {
|
|
|
|
| 497 |
"csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless8-2022-11-14": _get_librispeech_pre_trained_model, # noqa
|
| 498 |
"csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless7-2022-11-11": _get_librispeech_pre_trained_model, # noqa
|
| 499 |
"csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13": _get_librispeech_pre_trained_model, # noqa
|
| 500 |
+
"csukuangfj/wenet-english-model": _get_wenet_model,
|
| 501 |
}
|
| 502 |
|
| 503 |
chinese_english_mixed_models = {
|