Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
from pathlib import Path
import subprocess
import sys
from zipfile import ZipFile
ROOT = Path(__file__).resolve().parents[1]
RUNTIME_PACKAGE_FILES = {
"funasr/models/sense_voice/whisper_lib/normalizers/english.json",
"funasr/models/rwkv_bat/cuda_encoder/wkv_cuda.cu",
"funasr/models/rwkv_bat/cuda_encoder/wkv_op.cpp",
"funasr/models/rwkv_bat/cuda_decoder/wkv_cuda.cu",
"funasr/models/rwkv_bat/cuda_decoder/wkv_op.cpp",
}
def test_wheel_includes_runtime_package_data(tmp_path):
subprocess.run(
[
sys.executable,
"-c",
"import sys; from setuptools import build_meta; "
"build_meta.build_wheel(sys.argv[1])",
str(tmp_path),
],
check=True,
cwd=ROOT,
)
[wheel_path] = tmp_path.glob("*.whl")
with ZipFile(wheel_path) as wheel:
wheel_files = set(wheel.namelist())
assert RUNTIME_PACKAGE_FILES <= wheel_files