Spaces:
Sleeping
Sleeping
Rename app (2).py to app.py
Browse files- app (2).py β app.py +8 -25
app (2).py β app.py
RENAMED
@@ -45,10 +45,9 @@ This Gradio demo showcases **IndicConformer**, a speech recognition model for **
|
|
45 |
#### **How to Use:**
|
46 |
1. **Upload or record** an audio clip in any supported Indian language.
|
47 |
2. Select the **mode** (CTC or RNNT) for transcription.
|
48 |
-
3. Click **"Transcribe"** to generate the corresponding text
|
49 |
4. View or copy the output for further use.
|
50 |
|
51 |
-
π Try it out and experience seamless speech recognition for Indian languages!
|
52 |
"""
|
53 |
|
54 |
hf_token = os.getenv("HF_TOKEN")
|
@@ -62,8 +61,8 @@ model.eval()
|
|
62 |
CACHE_EXAMPLES = os.getenv("CACHE_EXAMPLES") == "1" and torch.cuda.is_available()
|
63 |
|
64 |
AUDIO_SAMPLE_RATE = 16000
|
65 |
-
MAX_INPUT_AUDIO_LENGTH =
|
66 |
-
DEFAULT_TARGET_LANGUAGE = "
|
67 |
|
68 |
@spaces.GPU
|
69 |
def run_asr_ctc(input_audio: str, target_language: str) -> str:
|
@@ -156,14 +155,6 @@ with gr.Blocks() as demo_asr_ctc:
|
|
156 |
btn = gr.Button("Transcribe")
|
157 |
with gr.Column():
|
158 |
output_text = gr.Textbox(label="Transcribed text")
|
159 |
-
|
160 |
-
gr.Examples(
|
161 |
-
examples=[
|
162 |
-
["assets/Bengali.wav", "Bengali", "English"],
|
163 |
-
["assets/Gujarati.wav", "Gujarati", "Hindi"],
|
164 |
-
["assets/Punjabi.wav", "Punjabi", "Hindi"],
|
165 |
-
|
166 |
-
],
|
167 |
inputs=[input_audio, target_language],
|
168 |
outputs=output_text,
|
169 |
fn=run_asr_ctc,
|
@@ -191,14 +182,6 @@ with gr.Blocks() as demo_asr_rnnt:
|
|
191 |
btn = gr.Button("Transcribe")
|
192 |
with gr.Column():
|
193 |
output_text = gr.Textbox(label="Transcribed text")
|
194 |
-
|
195 |
-
gr.Examples(
|
196 |
-
examples=[
|
197 |
-
["assets/Bengali.wav", "Bengali", "English"],
|
198 |
-
["assets/Gujarati.wav", "Gujarati", "Hindi"],
|
199 |
-
["assets/Punjabi.wav", "Punjabi", "Hindi"],
|
200 |
-
|
201 |
-
],
|
202 |
inputs=[input_audio, target_language],
|
203 |
outputs=output_text,
|
204 |
fn=run_asr_rnnt,
|
@@ -216,11 +199,11 @@ with gr.Blocks() as demo_asr_rnnt:
|
|
216 |
|
217 |
with gr.Blocks(css="style.css") as demo:
|
218 |
gr.Markdown(DESCRIPTION)
|
219 |
-
gr.DuplicateButton(
|
220 |
-
value="Duplicate Space for private use",
|
221 |
-
elem_id="duplicate-button",
|
222 |
-
visible=os.getenv("SHOW_DUPLICATE_BUTTON") == "1",
|
223 |
-
)
|
224 |
|
225 |
with gr.Tabs():
|
226 |
with gr.Tab(label="CTC"):
|
|
|
45 |
#### **How to Use:**
|
46 |
1. **Upload or record** an audio clip in any supported Indian language.
|
47 |
2. Select the **mode** (CTC or RNNT) for transcription.
|
48 |
+
3. Click **"Transcribe"** to generate the corresponding text.
|
49 |
4. View or copy the output for further use.
|
50 |
|
|
|
51 |
"""
|
52 |
|
53 |
hf_token = os.getenv("HF_TOKEN")
|
|
|
61 |
CACHE_EXAMPLES = os.getenv("CACHE_EXAMPLES") == "1" and torch.cuda.is_available()
|
62 |
|
63 |
AUDIO_SAMPLE_RATE = 16000
|
64 |
+
MAX_INPUT_AUDIO_LENGTH = 600 # in seconds
|
65 |
+
DEFAULT_TARGET_LANGUAGE = "Hindi"
|
66 |
|
67 |
@spaces.GPU
|
68 |
def run_asr_ctc(input_audio: str, target_language: str) -> str:
|
|
|
155 |
btn = gr.Button("Transcribe")
|
156 |
with gr.Column():
|
157 |
output_text = gr.Textbox(label="Transcribed text")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
158 |
inputs=[input_audio, target_language],
|
159 |
outputs=output_text,
|
160 |
fn=run_asr_ctc,
|
|
|
182 |
btn = gr.Button("Transcribe")
|
183 |
with gr.Column():
|
184 |
output_text = gr.Textbox(label="Transcribed text")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
185 |
inputs=[input_audio, target_language],
|
186 |
outputs=output_text,
|
187 |
fn=run_asr_rnnt,
|
|
|
199 |
|
200 |
with gr.Blocks(css="style.css") as demo:
|
201 |
gr.Markdown(DESCRIPTION)
|
202 |
+
# gr.DuplicateButton(
|
203 |
+
# value="Duplicate Space for private use",
|
204 |
+
# elem_id="duplicate-button",
|
205 |
+
# visible=os.getenv("SHOW_DUPLICATE_BUTTON") == "1",
|
206 |
+
# )
|
207 |
|
208 |
with gr.Tabs():
|
209 |
with gr.Tab(label="CTC"):
|