Spaces:
Running
on
Zero
Running
on
Zero
Commit
Β·
18b21ee
1
Parent(s):
e888ead
add actual code
Browse files- app.py +111 -8
- requirements.txt +6 -0
app.py
CHANGED
@@ -1,14 +1,117 @@
|
|
|
|
1 |
import gradio as gr
|
2 |
-
import spaces
|
3 |
import torch
|
|
|
|
|
|
|
|
|
4 |
|
5 |
-
|
6 |
-
print(
|
|
|
|
|
7 |
|
|
|
8 |
@spaces.GPU
|
9 |
-
def
|
10 |
-
print(
|
11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
-
|
14 |
-
|
|
|
|
1 |
+
import os
|
2 |
import gradio as gr
|
|
|
3 |
import torch
|
4 |
+
import spaces
|
5 |
+
import nemo.collections.asr as nemo_asr
|
6 |
+
from omegaconf import OmegaConf
|
7 |
+
import time
|
8 |
|
9 |
+
# Check if CUDA is available
|
10 |
+
print(f"CUDA available: {torch.cuda.is_available()}")
|
11 |
+
if torch.cuda.is_available():
|
12 |
+
print(f"CUDA device: {torch.cuda.get_device_name(0)}")
|
13 |
|
14 |
+
# Initialize the ASR model
|
15 |
@spaces.GPU
|
16 |
+
def load_model():
|
17 |
+
print("Loading ASR model...")
|
18 |
+
# Load the NVIDIA NeMo ASR model
|
19 |
+
model = nemo_asr.models.EncDecRNNTBPEModel.from_pretrained("nvidia/parakeet-tdt-0.6b-v2")
|
20 |
+
# Move model to GPU if available
|
21 |
+
if torch.cuda.is_available():
|
22 |
+
model = model.cuda()
|
23 |
+
print(f"Model loaded on device: {model.device}")
|
24 |
+
return model
|
25 |
+
|
26 |
+
# Global variable to store the model
|
27 |
+
model = load_model()
|
28 |
+
|
29 |
+
def transcribe(audio, state=""):
|
30 |
+
"""
|
31 |
+
Transcribe audio in real-time
|
32 |
+
"""
|
33 |
+
# Skip processing if no audio is provided
|
34 |
+
if audio is None:
|
35 |
+
return state, state
|
36 |
+
|
37 |
+
# Get the sample rate from the audio
|
38 |
+
sample_rate = 16000 # Default to 16kHz if not specified
|
39 |
+
|
40 |
+
# Process the audio with the ASR model
|
41 |
+
with torch.no_grad():
|
42 |
+
transcription = model.transcribe([audio])[0]
|
43 |
+
|
44 |
+
# Append new transcription to the state
|
45 |
+
if state == "":
|
46 |
+
new_state = transcription
|
47 |
+
else:
|
48 |
+
new_state = state + " " + transcription
|
49 |
+
|
50 |
+
return new_state, new_state
|
51 |
+
|
52 |
+
# Define the Gradio interface
|
53 |
+
with gr.Blocks(title="Real-time Speech-to-Text with NeMo") as demo:
|
54 |
+
gr.Markdown("# ποΈ Real-time Speech-to-Text Transcription")
|
55 |
+
gr.Markdown("Powered by NVIDIA NeMo and the parakeet-tdt-0.6b-v2 model")
|
56 |
+
|
57 |
+
with gr.Row():
|
58 |
+
with gr.Column(scale=2):
|
59 |
+
audio_input = gr.Audio(
|
60 |
+
source="microphone",
|
61 |
+
type="numpy",
|
62 |
+
streaming=True,
|
63 |
+
label="Speak into your microphone"
|
64 |
+
)
|
65 |
+
|
66 |
+
clear_btn = gr.Button("Clear Transcript")
|
67 |
+
|
68 |
+
with gr.Column(scale=3):
|
69 |
+
text_output = gr.Textbox(
|
70 |
+
label="Transcription",
|
71 |
+
placeholder="Your speech will appear here...",
|
72 |
+
lines=10
|
73 |
+
)
|
74 |
+
streaming_text = gr.Textbox(
|
75 |
+
label="Real-time Transcription",
|
76 |
+
placeholder="Real-time results will appear here...",
|
77 |
+
lines=2
|
78 |
+
)
|
79 |
+
|
80 |
+
# State to store the ongoing transcription
|
81 |
+
state = gr.State("")
|
82 |
+
|
83 |
+
# Handle the audio stream
|
84 |
+
audio_input.stream(
|
85 |
+
fn=transcribe,
|
86 |
+
inputs=[audio_input, state],
|
87 |
+
outputs=[state, streaming_text],
|
88 |
+
)
|
89 |
+
|
90 |
+
# Clear the transcription
|
91 |
+
def clear_transcription():
|
92 |
+
return "", "", ""
|
93 |
+
|
94 |
+
clear_btn.click(
|
95 |
+
fn=clear_transcription,
|
96 |
+
inputs=[],
|
97 |
+
outputs=[text_output, streaming_text, state]
|
98 |
+
)
|
99 |
+
|
100 |
+
# Update the main text output when the state changes
|
101 |
+
state.change(
|
102 |
+
fn=lambda s: s,
|
103 |
+
inputs=[state],
|
104 |
+
outputs=[text_output]
|
105 |
+
)
|
106 |
+
|
107 |
+
gr.Markdown("## π Instructions")
|
108 |
+
gr.Markdown("""
|
109 |
+
1. Click the microphone button to start recording
|
110 |
+
2. Speak clearly into your microphone
|
111 |
+
3. The transcription will appear in real-time
|
112 |
+
4. Click 'Clear Transcript' to start a new transcription
|
113 |
+
""")
|
114 |
|
115 |
+
# Launch the app
|
116 |
+
if __name__ == "__main__":
|
117 |
+
demo.launch()
|
requirements.txt
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
torch>=1.13.0
|
2 |
+
gradio>=3.32.0
|
3 |
+
nemo_toolkit[asr]>=1.18.0
|
4 |
+
omegaconf>=2.2.0
|
5 |
+
spaces>=0.15.0
|
6 |
+
numpy>=1.22.0
|