Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions .github/workflows/test-numpy-compatibility.yml
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,7 @@ on:
- "funasr/cli.py"
- "tests/test_cli.py"
- "funasr/auto/auto_model.py"
- "tests/test_auto_model_zero_speech_time.py"
- "funasr/tokenizer/sentencepiece_tokenizer.py"
- "funasr/tokenizer/char_tokenizer.py"
- "funasr/tokenizer/phoneme_tokenizer.py"
Expand Down Expand Up @@ -54,6 +55,7 @@ on:
- "funasr/cli.py"
- "tests/test_cli.py"
- "funasr/auto/auto_model.py"
- "tests/test_auto_model_zero_speech_time.py"
- "funasr/tokenizer/sentencepiece_tokenizer.py"
- "funasr/tokenizer/char_tokenizer.py"
- "funasr/tokenizer/phoneme_tokenizer.py"
Expand Down Expand Up @@ -158,6 +160,7 @@ jobs:
tests/test_paraformer_hotword_file.py \
tests/test_ct_transformer_streaming_empty_input.py \
tests/test_cli.py \
tests/test_auto_model_zero_speech_time.py \
--junitxml=numpy-results.xml
- uses: actions/upload-artifact@v4
if: always()
Expand Down
7 changes: 5 additions & 2 deletions funasr/auto/auto_model.py
Original file line number Diff line number Diff line change
Expand Up @@ -983,7 +983,9 @@ def inference(
speed_stats["extract_feat"] = meta_data.get("extract_feat", 0.0)
speed_stats["forward"] = f"{time_escape:0.3f}"
speed_stats["batch_size"] = f"{len(results)}"
speed_stats["rtf"] = f"{(time_escape) / batch_data_time:0.3f}"
# Input shorter than one feature frame reports zero speech time.
rtf = time_escape / batch_data_time if batch_data_time else float("nan")
speed_stats["rtf"] = f"{rtf:0.3f}"
description = f"{speed_stats}, "
if pbar:
pbar.update(end_idx - beg_idx)
Expand All @@ -998,7 +1000,8 @@ def inference(

if pbar:
# pbar.update(1)
pbar.set_description(f"rtf_avg: {time_escape_total/time_speech_total:0.3f}")
rtf_avg = time_escape_total / time_speech_total if time_speech_total else float("nan")
pbar.set_description(f"rtf_avg: {rtf_avg:0.3f}")

device = next(model.parameters()).device
if device.type == "cuda":
Expand Down
39 changes: 39 additions & 0 deletions tests/test_auto_model_zero_speech_time.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
import unittest

import torch

from funasr.auto.auto_model import AutoModel


class _ZeroSpeechTimeModel:
"""Reports no speech time, as models do for input shorter than one feature frame."""

def __init__(self):
self.param = torch.nn.Parameter(torch.zeros(1))

def parameters(self):
return iter([self.param])

def inference(self, data_in, **kwargs):
return [{"key": d, "value": []} for d in data_in], {"batch_data_time": 0.0}


def _auto_model(disable_pbar):
am = AutoModel.__new__(AutoModel)
am.model = _ZeroSpeechTimeModel()
am.kwargs = {"batch_size": 1, "disable_pbar": disable_pbar}
return am


class TestInferenceZeroSpeechTime(unittest.TestCase):
def test_zero_speech_time_returns_results_without_progress_bar(self):
res = AutoModel.inference(_auto_model(True), ["a", "b"])
self.assertEqual(res, [{"key": "a", "value": []}, {"key": "b", "value": []}])

def test_zero_speech_time_returns_results_with_progress_bar(self):
res = AutoModel.inference(_auto_model(False), ["a"])
self.assertEqual(res, [{"key": "a", "value": []}])


if __name__ == "__main__":
unittest.main()
Loading