Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,94 +1,103 @@
|
|
1 |
import gradio as gr
|
2 |
-
import
|
3 |
import matplotlib.pyplot as plt
|
4 |
-
import plotly.express as px
|
5 |
-
from radar_chart import radar_factory
|
6 |
-
|
7 |
-
from keras.models import load_model
|
8 |
-
import os
|
9 |
import numpy as np
|
|
|
|
|
|
|
|
|
10 |
|
|
|
11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
-
|
|
|
14 |
|
|
|
|
|
|
|
15 |
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
"""
|
20 |
-
|
21 |
-
# label_conversion = {0: 'neutral',
|
22 |
-
# 1: 'calm',
|
23 |
-
# 2: 'happy',
|
24 |
-
# 3: 'sad',
|
25 |
-
# 4: 'angry',
|
26 |
-
# 5: 'fearful',
|
27 |
-
# 6: 'disgust',
|
28 |
-
# 7: 'surprised'}
|
29 |
|
30 |
-
|
31 |
-
1: 'kata_benda',
|
32 |
-
2: 'kata_kerja',
|
33 |
-
3: 'kata_keterangan}
|
34 |
|
35 |
-
|
36 |
|
|
|
37 |
|
38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
"""
|
40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
"""
|
42 |
-
|
43 |
-
|
44 |
-
elif file is None and micro is not None:
|
45 |
-
input_audio = micro
|
46 |
-
else:
|
47 |
-
print("THERE IS A PROBLEM")
|
48 |
-
input_audio = file
|
49 |
-
|
50 |
-
data, sampling_rate = librosa.load(input_audio)
|
51 |
-
print(data)
|
52 |
-
print(f"THE SAMPLING RATE IS {sampling_rate}")
|
53 |
-
mfccs = np.mean(librosa.feature.mfcc(y=data, sr=sampling_rate, n_mfcc=40).T, axis=0)
|
54 |
-
x = np.expand_dims(mfccs, axis=1)
|
55 |
-
x = np.expand_dims(x, axis=0)
|
56 |
-
predictions = np.argmax(model.predict(x), axis=1)
|
57 |
-
|
58 |
-
N = 8
|
59 |
-
theta = radar_factory(N, frame='polygon')
|
60 |
-
spoke_labels = np.array(['kata_benda',
|
61 |
-
'kata_kerja',
|
62 |
-
'kata_keterangan',
|
63 |
-
'kata_sifat'])
|
64 |
-
fig_radar, axs = plt.subplots(figsize=(8, 8), nrows=1, ncols=1,
|
65 |
-
subplot_kw=dict(projection='radar'))
|
66 |
-
vec = model.predict(x)[0]
|
67 |
-
axs.plot(theta, vec, color="b")
|
68 |
-
axs.fill(theta, vec, alpha=0.3)
|
69 |
-
|
70 |
-
axs.set_varlabels(spoke_labels)
|
71 |
-
|
72 |
-
fig = plt.figure()
|
73 |
-
plt.plot(data, alpha=0.8)
|
74 |
-
plt.xlabel("temps")
|
75 |
-
plt.ylabel("amplitude")
|
76 |
-
|
77 |
-
|
78 |
-
return convert_class_to_emotion(predictions), fig, fig_radar
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
# Set the starting state to an empty string
|
83 |
-
iface = gr.Interface(
|
84 |
-
fn=make_predictions,
|
85 |
-
title="identify emotion of a chunk of audio speech",
|
86 |
-
description="a simple interface to perform emotion recognition from an audio file",
|
87 |
-
article="Author: <a href=\"https://huggingface.co/poisso\">Poisso</a>.",
|
88 |
-
inputs=[gr.Audio(source="upload", type="filepath", label="File"),
|
89 |
-
gr.Audio(source="microphone", type="filepath", streaming=False, label="Microphone")]
|
90 |
-
,
|
91 |
-
examples=[[os.path.join("examples", filename)] for filename in os.listdir("examples")],
|
92 |
-
outputs=[gr.Textbox(label="Text output"), gr.Plot(), gr.Plot()]
|
93 |
)
|
94 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
+
import wave
|
3 |
import matplotlib.pyplot as plt
|
|
|
|
|
|
|
|
|
|
|
4 |
import numpy as np
|
5 |
+
from extract_features import *
|
6 |
+
import pickle
|
7 |
+
import soundfile
|
8 |
+
import librosa
|
9 |
|
10 |
+
classifier = pickle.load(open('finalized_rf.sav', 'rb'))
|
11 |
|
12 |
+
def emotion_predict(input):
|
13 |
+
input_features = extract_feature(input, mfcc=True, chroma=True, mel=True, contrast=True, tonnetz=True)
|
14 |
+
rf_prediction = classifier.predict(input_features.reshape(1,-1))
|
15 |
+
if rf_prediction == 'happy':
|
16 |
+
return 'Happy π'
|
17 |
+
elif rf_prediction == 'neutral':
|
18 |
+
return 'Neutral π'
|
19 |
+
elif rf_prediction == 'sad':
|
20 |
+
return 'Sad π’'
|
21 |
+
else:
|
22 |
+
return 'Angry π€'
|
23 |
+
|
24 |
|
25 |
+
def plot_fig(input):
|
26 |
+
wav = wave.open(input, 'r')
|
27 |
|
28 |
+
raw = wav.readframes(-1)
|
29 |
+
raw = np.frombuffer(raw, "int16")
|
30 |
+
sampleRate = wav.getframerate()
|
31 |
|
32 |
+
Time = np.linspace(0, len(raw)/sampleRate, num=len(raw))
|
33 |
+
|
34 |
+
fig = plt.figure()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
35 |
|
36 |
+
plt.rcParams["figure.figsize"] = (50,15)
|
|
|
|
|
|
|
37 |
|
38 |
+
plt.title("Waveform Of the Audio", fontsize=25)
|
39 |
|
40 |
+
plt.xticks(fontsize=15)
|
41 |
|
42 |
+
plt.yticks(fontsize=15)
|
43 |
+
|
44 |
+
plt.ylabel("Amplitude", fontsize=25)
|
45 |
+
|
46 |
+
plt.plot(Time, raw, color='red')
|
47 |
+
|
48 |
+
return fig
|
49 |
+
|
50 |
+
|
51 |
+
with gr.Blocks() as app:
|
52 |
+
gr.Markdown(
|
53 |
"""
|
54 |
+
# Speech Emotion Detector π΅π
|
55 |
+
This application classifies inputted audio π according to the verbal emotion into four categories:
|
56 |
+
1. Happy π
|
57 |
+
2. Neutral π
|
58 |
+
3. Sad π’
|
59 |
+
4. Angry π€
|
60 |
+
"""
|
61 |
+
)
|
62 |
+
with gr.Tab("Record Audio"):
|
63 |
+
record_input = gr.Audio(source="microphone", type="filepath")
|
64 |
+
|
65 |
+
with gr.Accordion("Audio Visualization", open=False):
|
66 |
+
gr.Markdown(
|
67 |
+
"""
|
68 |
+
### Visualization will work only after Audio has been submitted
|
69 |
+
"""
|
70 |
+
)
|
71 |
+
plot_record = gr.Button("Display Audio Signal")
|
72 |
+
plot_record_c = gr.Plot(label='Waveform Of the Audio')
|
73 |
+
|
74 |
+
record_button = gr.Button("Detect Emotion")
|
75 |
+
record_output = gr.Text(label = 'Emotion Detected')
|
76 |
+
|
77 |
+
with gr.Tab("Upload Audio File"):
|
78 |
+
gr.Markdown(
|
79 |
"""
|
80 |
+
## Uploaded Audio should be of .wav format
|
81 |
+
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
82 |
)
|
83 |
+
|
84 |
+
upload_input = gr.Audio(type="filepath")
|
85 |
+
|
86 |
+
with gr.Accordion("Audio Visualization", open=False):
|
87 |
+
gr.Markdown(
|
88 |
+
"""
|
89 |
+
### Visualization will work only after Audio has been submitted
|
90 |
+
"""
|
91 |
+
)
|
92 |
+
plot_upload = gr.Button("Display Audio Signal")
|
93 |
+
plot_upload_c = gr.Plot(label='Waveform Of the Audio')
|
94 |
+
|
95 |
+
upload_button = gr.Button("Detect Emotion")
|
96 |
+
upload_output = gr.Text(label = 'Emotion Detected')
|
97 |
+
|
98 |
+
record_button.click(emotion_predict, inputs=record_input, outputs=record_output)
|
99 |
+
upload_button.click(emotion_predict, inputs=upload_input, outputs=upload_output)
|
100 |
+
plot_record.click(plot_fig, inputs=record_input, outputs=plot_record_c)
|
101 |
+
plot_upload.click(plot_fig, inputs=upload_input, outputs=plot_upload_c)
|
102 |
+
|
103 |
+
app.launch()
|