EngineX-Iluvatar/enginex_bi_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

Replace torchaudio with soundfile in python-api-examples (#765)

Browse Source
This commit is contained in:
gtf35
2024-04-13 23:39:07 +08:00
committed by GitHub
parent 983df28a83
commit b0265b258d
3 changed files with 27 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
python-api-examples/speaker-identification-with-vad-non-streaming-asr.py
View File

@@ -65,7 +65,7 @@ from typing import Dict, List, Tuple
import numpy as np import numpy as np
import sherpa_onnx import sherpa_onnx
import torchaudio import soundfile as sf
try: try:
import sounddevice as sd import sounddevice as sd
@@ -357,8 +357,14 @@ def load_speaker_file(args) -> Dict[str, List[str]]:
def load_audio(filename: str) -> Tuple[np.ndarray, int]: def load_audio(filename: str) -> Tuple[np.ndarray, int]:
samples, sample_rate = torchaudio.load(filename) data, sample_rate = sf.read(
return samples[0].contiguous().numpy(), sample_rate filename,
always_2d=True,
dtype="float32",
)
data = data[:, 0] # use only the first channel
samples = np.ascontiguousarray(data)
return samples, sample_rate
def compute_speaker_embedding( def compute_speaker_embedding(

12
python-api-examples/speaker-identification-with-vad.py
View File

@@ -60,7 +60,7 @@ from typing import Dict, List, Tuple
import numpy as np import numpy as np
import sherpa_onnx import sherpa_onnx
import torchaudio import soundfile as sf
try: try:
import sounddevice as sd import sounddevice as sd
@@ -160,8 +160,14 @@ def load_speaker_file(args) -> Dict[str, List[str]]:
def load_audio(filename: str) -> Tuple[np.ndarray, int]: def load_audio(filename: str) -> Tuple[np.ndarray, int]:
samples, sample_rate = torchaudio.load(filename) data, sample_rate = sf.read(
return samples[0].contiguous().numpy(), sample_rate filename,
always_2d=True,
dtype="float32",
)
data = data[:, 0] # use only the first channel
samples = np.ascontiguousarray(data)
return samples, sample_rate
def compute_speaker_embedding( def compute_speaker_embedding(

12
python-api-examples/speaker-identification.py
View File

@@ -52,7 +52,7 @@ from typing import Dict, List, Tuple
import numpy as np import numpy as np
import sherpa_onnx import sherpa_onnx
import torchaudio import soundfile as sf
try: try:
import sounddevice as sd import sounddevice as sd
@@ -145,8 +145,14 @@ def load_speaker_file(args) -> Dict[str, List[str]]:
def load_audio(filename: str) -> Tuple[np.ndarray, int]: def load_audio(filename: str) -> Tuple[np.ndarray, int]:
samples, sample_rate = torchaudio.load(filename) data, sample_rate = sf.read(
return samples[0].contiguous().numpy(), sample_rate filename,
always_2d=True,
dtype="float32",
)
data = data[:, 0] # use only the first channel
samples = np.ascontiguousarray(data)
return samples, sample_rate
def compute_speaker_embedding( def compute_speaker_embedding(