Spaces:
Running
Running
Fix typo
Browse files
app.py
CHANGED
@@ -11,12 +11,13 @@ import audiofile
|
|
11 |
import audresample
|
12 |
|
13 |
|
|
|
14 |
model_name = "audeering/wav2vec2-large-robust-24-ft-age-gender"
|
15 |
duration = 1 # limit processing of audio
|
16 |
|
17 |
|
18 |
-
class
|
19 |
-
r"""
|
20 |
|
21 |
def __init__(self, config, num_labels):
|
22 |
|
@@ -39,7 +40,7 @@ class ModelHead(nn.Module):
|
|
39 |
|
40 |
|
41 |
class AgeGenderModel(Wav2Vec2PreTrainedModel):
|
42 |
-
r"""
|
43 |
|
44 |
def __init__(self, config):
|
45 |
|
@@ -47,8 +48,8 @@ class AgeGenderModel(Wav2Vec2PreTrainedModel):
|
|
47 |
|
48 |
self.config = config
|
49 |
self.wav2vec2 = Wav2Vec2Model(config)
|
50 |
-
self.age =
|
51 |
-
self.gender =
|
52 |
self.init_weights()
|
53 |
|
54 |
def forward(
|
@@ -67,7 +68,6 @@ class AgeGenderModel(Wav2Vec2PreTrainedModel):
|
|
67 |
|
68 |
|
69 |
# load model from hub
|
70 |
-
device = 0 if torch.cuda.is_available() else "cpu"
|
71 |
processor = Wav2Vec2Processor.from_pretrained(model_name)
|
72 |
model = AgeGenderModel.from_pretrained(model_name)
|
73 |
|
@@ -105,14 +105,14 @@ def process_func(x: np.ndarray, sampling_rate: int) -> dict:
|
|
105 |
def recognize(input_file):
|
106 |
# sampling_rate, signal = input_microphone
|
107 |
# signal = signal.astype(np.float32, order="C") / 32768.0
|
108 |
-
if
|
109 |
-
signal, sampling_rate = audiofile.read(input_file, duration=duration)
|
110 |
-
else:
|
111 |
raise gr.Error(
|
112 |
"No audio file submitted! "
|
113 |
"Please upload or record an audio file "
|
114 |
"before submitting your request."
|
115 |
)
|
|
|
|
|
116 |
# Resample to sampling rate supported byu the models
|
117 |
target_rate = 16000
|
118 |
signal = audresample.resample(signal, sampling_rate, target_rate)
|
|
|
11 |
import audresample
|
12 |
|
13 |
|
14 |
+
device = 0 if torch.cuda.is_available() else "cpu"
|
15 |
model_name = "audeering/wav2vec2-large-robust-24-ft-age-gender"
|
16 |
duration = 1 # limit processing of audio
|
17 |
|
18 |
|
19 |
+
class AgeGenderHead(nn.Module):
|
20 |
+
r"""Age-gender model head."""
|
21 |
|
22 |
def __init__(self, config, num_labels):
|
23 |
|
|
|
40 |
|
41 |
|
42 |
class AgeGenderModel(Wav2Vec2PreTrainedModel):
|
43 |
+
r"""Age-gender recognition model."""
|
44 |
|
45 |
def __init__(self, config):
|
46 |
|
|
|
48 |
|
49 |
self.config = config
|
50 |
self.wav2vec2 = Wav2Vec2Model(config)
|
51 |
+
self.age = AgeGenderHead(config, 1)
|
52 |
+
self.gender = AgeGenderHead(config, 3)
|
53 |
self.init_weights()
|
54 |
|
55 |
def forward(
|
|
|
68 |
|
69 |
|
70 |
# load model from hub
|
|
|
71 |
processor = Wav2Vec2Processor.from_pretrained(model_name)
|
72 |
model = AgeGenderModel.from_pretrained(model_name)
|
73 |
|
|
|
105 |
def recognize(input_file):
|
106 |
# sampling_rate, signal = input_microphone
|
107 |
# signal = signal.astype(np.float32, order="C") / 32768.0
|
108 |
+
if input_file is None:
|
|
|
|
|
109 |
raise gr.Error(
|
110 |
"No audio file submitted! "
|
111 |
"Please upload or record an audio file "
|
112 |
"before submitting your request."
|
113 |
)
|
114 |
+
|
115 |
+
signal, sampling_rate = audiofile.read(input_file, duration=duration)
|
116 |
# Resample to sampling rate supported byu the models
|
117 |
target_rate = 16000
|
118 |
signal = audresample.resample(signal, sampling_rate, target_rate)
|