Initial commit: FunASR Speech Recognition Toolkit

Add complete FunASR codebase including models, runtime, and documentation.
2026-07-09 22:38:58 +08:00
commit 6116b1f3c6
3683 changed files with 990984 additions and 0 deletions
@@ -0,0 +1,43 @@
+from funasr.register import tables
+from funasr.models.whisper_lid.eres2net.ResNet import (
+    ERes2Net,
+    BasicBlockERes2Net,
+    BasicBlockERes2Net_diff_AFF,
+)
+
+
+@tables.register("lid_predictor_classes", "LidPredictor")
+class LidPredictor(ERes2Net):
+    def __init__(
+        self,
+        block=BasicBlockERes2Net,
+        block_fuse=BasicBlockERes2Net_diff_AFF,
+        num_blocks=[3, 4, 6, 3],
+        m_channels=32,
+        feat_dim=80,
+        embedding_size=192,
+        pooling_func="TSTP",
+        two_emb_layer=False,
+    ):
+        """Initialize LidPredictor.
+        
+            Args:
+                block: TODO.
+                block_fuse: TODO.
+                num_blocks: TODO.
+                m_channels: TODO.
+                feat_dim: Size/dimension parameter.
+                embedding_size: Size/dimension parameter.
+                pooling_func: TODO.
+                two_emb_layer: TODO.
+            """
+        super(LidPredictor, self).__init__(
+            block=block,
+            block_fuse=block_fuse,
+            num_blocks=num_blocks,
+            m_channels=m_channels,
+            feat_dim=feat_dim,
+            embedding_size=embedding_size,
+            pooling_func=pooling_func,
+            two_emb_layer=two_emb_layer,
+        )