Commit more scripts for wenetspeech kws recipe

k2-fsa · Feb 2, 2024 · 8b65f41 · 8b65f41
1 parent 4b33563
commit 8b65f41
Show file tree

Hide file tree

Showing 10 changed files with 2,353 additions and 147 deletions.
diff --git a/egs/gigaspeech/KWS/zipformer/finetune.py b/egs/gigaspeech/KWS/zipformer/finetune.py
@@ -158,6 +158,13 @@ def get_parser():
         formatter_class=argparse.ArgumentDefaultsHelpFormatter
     )
 
+    parser.add_argument(
+        "--bpe-model",
+        type=str,
+        default="data/lang_bpe_500/bpe.model",
+        help="Path to the BPE model",
+    )
+
     add_training_arguments(parser)
     add_model_arguments(parser)
     add_finetune_arguments(parser)

diff --git a/egs/wenetspeech/ASR/pruned_transducer_stateless2/asr_datamodule.py b/egs/wenetspeech/ASR/pruned_transducer_stateless2/asr_datamodule.py
@@ -407,10 +407,3 @@ def test_net_cuts(self) -> List[CutSet]:
     def test_meeting_cuts(self) -> List[CutSet]:
         logging.info("About to get TEST_MEETING cuts")
         return load_manifest_lazy(self.args.manifest_dir / "cuts_TEST_MEETING.jsonl.gz")
-
-    @lru_cache()
-    def test_open_commands_cuts(self) -> CutSet:
-        logging.info("About to get open commands cuts")
-        return load_manifest_lazy(
-            self.args.manifest_dir / "open-commands-cn_cuts_test.jsonl.gz"
-        )
diff --git a/egs/wenetspeech/KWS/shared b/egs/wenetspeech/KWS/shared
@@ -0,0 +1 @@
+../../../icefall/shared
diff --git a/egs/wenetspeech/KWS/zipformer/asr_datamodule.py b/egs/wenetspeech/KWS/zipformer/asr_datamodule.py
@@ -1,4 +1,5 @@
 # Copyright      2021  Piotr Żelasko
+# Copyright      2024  Xiaomi Corporation     (Author: Wei Kang)
 #
 # See ../../../../LICENSE for clarification regarding multiple authors
 #
@@ -409,8 +410,50 @@ def test_meeting_cuts(self) -> List[CutSet]:
         return load_manifest_lazy(self.args.manifest_dir / "cuts_TEST_MEETING.jsonl.gz")
 
     @lru_cache()
-    def test_open_commands_cuts(self) -> CutSet:
-        logging.info("About to get open commands cuts")
+    def cn_speech_commands_small_cuts(self) -> CutSet:
+        logging.info("About to get cn speech commands small cuts")
         return load_manifest_lazy(
-            self.args.manifest_dir / "open-commands-cn_cuts_test.jsonl.gz"
+            self.args.manifest_dir / "cn_speech_commands_cuts_small.jsonl.gz"
+        )
+
+    @lru_cache()
+    def cn_speech_commands_large_cuts(self) -> CutSet:
+        logging.info("About to get cn speech commands large cuts")
+        return load_manifest_lazy(
+            self.args.manifest_dir / "cn_speech_commands_cuts_large.jsonl.gz"
+        )
+
+    @lru_cache()
+    def nihaowenwen_dev_cuts(self) -> CutSet:
+        logging.info("About to get nihaowenwen dev cuts")
+        return load_manifest_lazy(
+            self.args.manifest_dir / "nihaowenwen_cuts_dev.jsonl.gz"
+        )
+
+    @lru_cache()
+    def nihaowenwen_test_cuts(self) -> CutSet:
+        logging.info("About to get nihaowenwen test cuts")
+        return load_manifest_lazy(
+            self.args.manifest_dir / "nihaowenwen_cuts_test.jsonl.gz"
+        )
+
+    @lru_cache()
+    def nihaowenwen_train_cuts(self) -> CutSet:
+        logging.info("About to get nihaowenwen train cuts")
+        return load_manifest_lazy(
+            self.args.manifest_dir / "nihaowenwen_cuts_train.jsonl.gz"
+        )
+
+    @lru_cache()
+    def xiaoyun_clean_cuts(self) -> CutSet:
+        logging.info("About to get xiaoyun clean cuts")
+        return load_manifest_lazy(
+            self.args.manifest_dir / "xiaoyun_cuts_clean.jsonl.gz"
+        )
+
+    @lru_cache()
+    def xiaoyun_noisy_cuts(self) -> CutSet:
+        logging.info("About to get xiaoyun noisy cuts")
+        return load_manifest_lazy(
+            self.args.manifest_dir / "xiaoyun_cuts_noisy.jsonl.gz"
         )