Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Reproduce results with latest librosa and torch #8

Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open
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
2 changes: 1 addition & 1 deletion datasets/speech_commands_dataset.py
Original file line number Diff line number Diff line change
Expand Up @@ -80,7 +80,7 @@ def __init__(self, folder, transform=None, sample_rate=16000, sample_length=1):
samples = []
for f in audio_files:
path = os.path.join(folder, f)
s, sr = librosa.load(path, sample_rate)
s, sr = librosa.load(path, sr=sample_rate)
samples.append(s)

samples = np.hstack(samples)
Expand Down
8 changes: 8 additions & 0 deletions requirements.txt
Original file line number Diff line number Diff line change
@@ -0,0 +1,8 @@
librosa==0.10.1
matplotlib==3.8.2
tensorboardX==2.6.2.2
torch==2.1.2
torchaudio==2.1.2
torchvision==0.16.2
tqdm==4.66.1
torchnet==0.0.4
6 changes: 3 additions & 3 deletions test_speech_commands.py
Original file line number Diff line number Diff line change
Expand Up @@ -89,12 +89,12 @@ def test():
if args.multi_crop:
inputs = multi_crop(inputs)

inputs = Variable(inputs, volatile = True)
inputs = Variable(inputs, requires_grad=False)
targets = Variable(targets, requires_grad=False)

if use_gpu:
inputs = inputs.cuda()
targets = targets.cuda(async=True)
targets = targets.cuda()

# forward
outputs = model(inputs)
Expand All @@ -111,7 +111,7 @@ def test():
pred = outputs.data.max(1, keepdim=True)[1]
correct += pred.eq(targets.data.view_as(pred)).sum()
total += targets.size(0)
confusion_matrix.add(pred, targets.data)
confusion_matrix.add(pred.squeeze(), targets.data)

filenames = batch['path']
for j in range(len(pred)):
Expand Down
14 changes: 7 additions & 7 deletions train_speech_commands.py
Original file line number Diff line number Diff line change
Expand Up @@ -152,7 +152,7 @@ def train(epoch):

if use_gpu:
inputs = inputs.cuda()
targets = targets.cuda(async=True)
targets = targets.cuda()

# forward/backward
outputs = model(inputs)
Expand All @@ -167,15 +167,15 @@ def train(epoch):
# statistics
it += 1
global_step += 1
running_loss += loss.data[0]
running_loss += loss.item()
pred = outputs.data.max(1, keepdim=True)[1]
if args.mixup:
targets = batch['target']
targets = Variable(targets, requires_grad=False).cuda(async=True)
targets = Variable(targets, requires_grad=False).cuda()
correct += pred.eq(targets.data.view_as(pred)).sum()
total += targets.size(0)

writer.add_scalar('%s/loss' % phase, loss.data[0], global_step)
writer.add_scalar('%s/loss' % phase, loss.item(), global_step)

# update the progress bar
pbar.set_postfix({
Expand Down Expand Up @@ -210,7 +210,7 @@ def valid(epoch):

if use_gpu:
inputs = inputs.cuda()
targets = targets.cuda(async=True)
targets = targets.cuda()

# forward
outputs = model(inputs)
Expand All @@ -219,12 +219,12 @@ def valid(epoch):
# statistics
it += 1
global_step += 1
running_loss += loss.data[0]
running_loss += loss.item()
pred = outputs.data.max(1, keepdim=True)[1]
correct += pred.eq(targets.data.view_as(pred)).sum()
total += targets.size(0)

writer.add_scalar('%s/loss' % phase, loss.data[0], global_step)
writer.add_scalar('%s/loss' % phase, loss.item(), global_step)

# update the progress bar
pbar.set_postfix({
Expand Down
4 changes: 2 additions & 2 deletions transforms/transforms_stft.py
Original file line number Diff line number Diff line change
Expand Up @@ -41,7 +41,7 @@ def __call__(self, data):
sample_rate = data['sample_rate']
hop_length = data['hop_length']
scale = random.uniform(-self.max_scale, self.max_scale)
stft_stretch = librosa.core.phase_vocoder(stft, 1+scale, hop_length=hop_length)
stft_stretch = librosa.core.phase_vocoder(stft, rate=1+scale, hop_length=hop_length)
data['stft'] = stft_stretch
return data

Expand Down Expand Up @@ -108,7 +108,7 @@ def __call__(self, data):
stft = data['stft']
sample_rate = data['sample_rate']
n_fft = data['n_fft']
mel_basis = librosa.filters.mel(sample_rate, n_fft, self.n_mels)
mel_basis = librosa.filters.mel(sr=sample_rate, n_fft=n_fft, n_mels=self.n_mels)
s = np.dot(mel_basis, np.abs(stft)**2.0)
data['mel_spectrogram'] = librosa.power_to_db(s, ref=np.max)
return data
Expand Down
4 changes: 2 additions & 2 deletions transforms/transforms_wav.py
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ def __init__(self, sample_rate=16000):
def __call__(self, data):
path = data['path']
if path:
samples, sample_rate = librosa.load(path, self.sample_rate)
samples, sample_rate = librosa.load(path, sr=self.sample_rate)
else:
# silence
sample_rate = self.sample_rate
Expand Down Expand Up @@ -137,7 +137,7 @@ def __init__(self, n_mels=32):
def __call__(self, data):
samples = data['samples']
sample_rate = data['sample_rate']
s = librosa.feature.melspectrogram(samples, sr=sample_rate, n_mels=self.n_mels)
s = librosa.feature.melspectrogram(y=samples, sr=sample_rate, n_mels=self.n_mels)
data['mel_spectrogram'] = librosa.power_to_db(s, ref=np.max)
return data

Expand Down