small fixes

app.py

@@ -125,9 +125,6 @@ def process(
 
     filename = convert_to_wav(in_filename)
 
-    logging.info(f"filename: {in_filename}")
-    os.system(f"ffprobe {filename}")
-
     now = datetime.now()
     date_time = now.strftime("%Y-%m-%d %H:%M:%S.%f")
     logging.info(f"Started at {date_time}")
@@ -319,6 +316,13 @@ with demo:
         )
     gr.Markdown(description)
 
+torch.set_num_threads(1)
+torch.set_num_interop_threads(1)
+
+torch._C._jit_set_profiling_executor(False)
+torch._C._jit_set_profiling_mode(False)
+torch._C._set_graph_executor_optimize(False)
+
 if __name__ == "__main__":
     formatter = "%(asctime)s %(levelname)s [%(filename)s:%(lineno)d] %(message)s"
 

examples.py

@@ -16,6 +16,48 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 examples = [
+    [
+        "Chinese+English",
+        "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh",
+        "greedy_search",
+        4,
+        "./test_wavs/tal_csasr/0.wav",
+    ],
+    [
+        "English",
+        "csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13",
+        "greedy_search",
+        4,
+        "./test_wavs/librispeech/1089-134686-0001.wav",
+    ],
+    [
+        "Chinese",
+        "luomingshuang/icefall_asr_wenetspeech_pruned_transducer_stateless2",
+        "greedy_search",
+        4,
+        "./test_wavs/wenetspeech/DEV_T0000000000.opus",
+    ],
+    [
+        "German",
+        "csukuangfj/wav2vec2.0-torchaudio",
+        "greedy_search",
+        4,
+        "./test_wavs/german/20170517-0900-PLENARY-16-de_20170517.wav",
+    ],
+    [
+        "Arabic",
+        "AmirHussein/icefall-asr-mgb2-conformer_ctc-2022-27-06",
+        "greedy_search",
+        4,
+        "./test_wavs/arabic/a.wav",
+    ],
+    [
+        "Tibetan",
+        "syzym/icefall-asr-xbmu-amdo31-pruned-transducer-stateless7-2022-12-02",
+        "greedy_search",
+        4,
+        "./test_wavs/tibetan/a_0_cacm-A70_31117.wav",
+    ],
     # librispeech
     # https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless5-2022-05-13/tree/main/test_wavs
     [
@@ -154,57 +196,42 @@ examples = [
         "./test_wavs/aidatatang_200zh/T0055G0036S0004.wav",
     ],
     # tal_csasr
-    # https://huggingface.co/luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5/tree/main/test_wavs
     [
         "Chinese+English",
-        "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5",
+        "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh",
         "greedy_search",
         4,
         "./test_wavs/tal_csasr/210_36476_210_8341_1_1533271973_7057520_132.wav",
     ],
     [
         "Chinese+English",
-        "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5",
+        "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh",
         "greedy_search",
         4,
         "./test_wavs/tal_csasr/210_36476_210_8341_1_1533271973_7057520_138.wav",
     ],
     [
         "Chinese+English",
-        "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5",
+        "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh",
         "greedy_search",
         4,
         "./test_wavs/tal_csasr/210_36476_210_8341_1_1533271973_7057520_145.wav",
     ],
     [
         "Tibetan",
-        "syzym/icefall-asr-xbmu-amdo31-pruned-transducer-stateless5-2022-11-29",
+        "syzym/icefall-asr-xbmu-amdo31-pruned-transducer-stateless7-2022-12-02",
         "greedy_search",
         4,
         "./test_wavs/tibetan/a_0_cacm-A70_31116.wav",
     ],
     [
         "Tibetan",
-        "syzym/icefall-asr-xbmu-amdo31-pruned-transducer-stateless5-2022-11-29",
-        "greedy_search",
-        4,
-        "./test_wavs/tibetan/a_0_cacm-A70_31117.wav",
-    ],
-    [
-        "Tibetan",
-        "syzym/icefall-asr-xbmu-amdo31-pruned-transducer-stateless5-2022-11-29",
+        "syzym/icefall-asr-xbmu-amdo31-pruned-transducer-stateless7-2022-12-02",
         "greedy_search",
         4,
         "./test_wavs/tibetan/a_0_cacm-A70_31118.wav",
     ],
     # arabic
-    [
-        "Arabic",
-        "AmirHussein/icefall-asr-mgb2-conformer_ctc-2022-27-06",
-        "greedy_search",
-        4,
-        "./test_wavs/arabic/a.wav",
-    ],
     [
         "Arabic",
         "AmirHussein/icefall-asr-mgb2-conformer_ctc-2022-27-06",
@@ -226,11 +253,4 @@ examples = [
         4,
         "./test_wavs/german/20120315-0900-PLENARY-14-de_20120315.wav",
     ],
-    [
-        "German",
-        "csukuangfj/wav2vec2.0-torchaudio",
-        "greedy_search",
-        4,
-        "./test_wavs/german/20170517-0900-PLENARY-16-de_20170517.wav",
-    ],
 ]

model.py

@@ -262,20 +262,28 @@ def _get_wenetspeech_pre_trained_model(
 
 
 @lru_cache(maxsize=10)
-def _get_tal_csasr_pre_trained_model(
+def _get_chinese_english_mixed_model(
     repo_id: str,
     decoding_method: str,
     num_active_paths: int,
 ):
     assert repo_id in [
         "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5",
+        "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh",
     ], repo_id
 
+    if repo_id == "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5":
+        filename = "cpu_jit.pt"
+        subfolder = "data/lang_char"
+    elif repo_id == "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh":
+        filename = "cpu_jit-epoch-11-avg-1.pt"
+        subfolder = "data/lang_char_bpe"
+
     nn_model = _get_nn_model_filename(
         repo_id=repo_id,
-        filename="cpu_jit.pt",
+        filename=filename,
     )
-    tokens = _get_token_filename(repo_id=repo_id)
+    tokens = _get_token_filename(repo_id=repo_id, subfolder=subfolder)
 
     feat_config = sherpa.FeatureConfig()
     feat_config.fbank_opts.frame_opts.samp_freq = sample_rate
@@ -541,7 +549,8 @@ english_models = {
 }
 
 chinese_english_mixed_models = {
-    "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5": _get_tal_csasr_pre_trained_model,  # noqa
+    "ptrnull/icefall-asr-conv-emformer-transducer-stateless2-zh": _get_chinese_english_mixed_model,
+    "luomingshuang/icefall_asr_tal-csasr_pruned_transducer_stateless5": _get_chinese_english_mixed_model,  # noqa
 }
 
 tibetan_models = {