Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

docs: complete model docstrings for transducer and whisper_lid

游雁77c72ee47589
diff --git a/funasr/models/transducer/model.py b/funasr/models/transducer/model.py
index e72200083..354413161 100644
--- a/funasr/models/transducer/model.py
+++ b/funasr/models/transducer/model.py
@@ -38,6 +38,14 @@ else:
@tables.register("model_classes", "Transducer")
class Transducer(torch.nn.Module):
+ """Transducer (RNN-T): Streaming ASR using encoder-predictor-joint architecture.
+
+ Combines encoder (audio frames), predictor (text history), and joint network.
+ Supports beam search decoding. Suitable for low-latency streaming applications.
+
+ Output: {"key": str, "text": str}
+ """
+
def __init__(
self,
frontend: Optional[str] = None,
diff --git a/funasr/models/whisper_lid/model.py b/funasr/models/whisper_lid/model.py
index baca008b0..e56c01c69 100644
--- a/funasr/models/whisper_lid/model.py
+++ b/funasr/models/whisper_lid/model.py
@@ -20,7 +20,13 @@ from funasr.register import tables
@tables.register("model_classes", "OpenAIWhisperModel")
class OpenAIWhisperModel(nn.Module):
- """CTC-attention hybrid Encoder-Decoder model"""
+ """Whisper-LID: Language Identification using OpenAI Whisper encoder.
+
+ Detects spoken language from audio input. Supports 99 languages.
+ Uses Whisper encoder features with a classification head.
+
+ Output: {"key": str, "text": str (language code)}
+ """
def __init__(
self,