update

toolbox/k2_sherpa/nn_models.py

@@ -312,8 +312,28 @@ model_map = {
             "tokens_file_sub_folder": "data/lang_bpe_500",
             "loader": "load_sherpa_offline_recognizer",
         },
+        {
+            "repo_id": "csukuangfj/wenet-english-model",
+            "nn_model_file": "final.zip",
+            "nn_model_file_sub_folder": ".",
+            "tokens_file": "units.txt",
+            "tokens_file_sub_folder": ".",
+            "loader": "load_sherpa_offline_recognizer",
+        },
     ],
     "Chinese+English": [
+        {
+            "repo_id": "csukuangfj/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20",
+            "encoder_model_file": "encoder-epoch-99-avg-1.onnx",
+            "encoder_model_file_sub_folder": ".",
+            "decoder_model_file": "decoder-epoch-99-avg-1.onnx",
+            "decoder_model_file_sub_folder": ".",
+            "joiner_model_file": "joiner-epoch-99-avg-1.onnx",
+            "joiner_model_file_sub_folder": ".",
+            "tokens_file": "tokens.txt",
+            "tokens_file_sub_folder": ".",
+            "loader": "load_sherpa_online_recognizer_from_transducer",
+        },
         {
             "repo_id": "csukuangfj/sherpa-onnx-paraformer-zh-2023-03-28",
             "nn_model_file": "model.int8.onnx",
@@ -322,6 +342,22 @@ model_map = {
             "tokens_file_sub_folder": ".",
             "loader": "load_sherpa_offline_recognizer_from_paraformer",
         },
+        {
+            "repo_id": "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh",
+            "nn_model_file": "cpu_jit-epoch-11-avg-1.pt",
+            "nn_model_file_sub_folder": "exp",
+            "tokens_file": "tokens.txt",
+            "tokens_file_sub_folder": "data/lang_char_bpe",
+            "loader": "load_sherpa_offline_recognizer",
+        },
+        {
+            "repo_id": "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5",
+            "nn_model_file": "cpu_jit.pt",
+            "nn_model_file_sub_folder": "exp",
+            "tokens_file": "tokens.txt",
+            "tokens_file_sub_folder": "data/lang_char",
+            "loader": "load_sherpa_offline_recognizer",
+        },
     ],
     "Chinese+Cantonese+English": [
         {
@@ -477,6 +513,30 @@ def load_sherpa_offline_recognizer_from_whisper(encoder_model_file: str,
     return recognizer
 
 
+def load_sherpa_online_recognizer_from_transducer(encoder_model_file: str,
+                                                  decoder_model_file: str,
+                                                  joiner_model_file: str,
+                                                  tokens_file: str,
+                                                  sample_rate: int = 16000,
+                                                  decoding_method: str = "greedy_search",
+                                                  feature_dim: int = 80,
+                                                  num_threads: int = 2,
+                                                  num_active_paths: int = 2,
+                                                  ):
+    recognizer = sherpa_onnx.OfflineRecognizer.from_transducer(
+        encoder=encoder_model_file,
+        decoder=decoder_model_file,
+        joiner=joiner_model_file,
+        tokens=tokens_file,
+        num_threads=num_threads,
+        sample_rate=sample_rate,
+        feature_dim=feature_dim,
+        decoding_method=decoding_method,
+        max_active_paths=num_active_paths,
+    )
+    return recognizer
+
+
 def load_recognizer(local_model_dir: Path,
                     decoding_method: str = "greedy_search",
                     num_active_paths: int = 4,
@@ -531,6 +591,10 @@ def load_recognizer(local_model_dir: Path,
         recognizer = load_sherpa_offline_recognizer_from_whisper(
             **kwargs_
         )
+    elif loader == "load_sherpa_online_recognizer_from_transducer":
+        recognizer = load_sherpa_online_recognizer_from_transducer(
+            **kwargs_
+        )
     else:
         raise NotImplementedError("loader not support: {}".format(loader))
     return recognizer