Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -8,19 +8,13 @@ import os
|
|
| 8 |
from transcribe_core import process_audio_file, get_audio_duration
|
| 9 |
from ai_providers import GeminiProvider, OpenRouterProvider
|
| 10 |
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
def transcribe_audio(audio_file, gemini_key, openrouter_key, model_name):
|
| 13 |
"""
|
| 14 |
Main transcription function for Gradio interface.
|
| 15 |
-
|
| 16 |
-
Args:
|
| 17 |
-
audio_file: Uploaded audio file
|
| 18 |
-
gemini_key: Gemini API key for transcription
|
| 19 |
-
openrouter_key: OpenRouter API key for summary/ideas
|
| 20 |
-
model_name: Gemini model to use
|
| 21 |
-
|
| 22 |
-
Returns:
|
| 23 |
-
Tuple of (status_message, download_file_path)
|
| 24 |
"""
|
| 25 |
if not audio_file:
|
| 26 |
return "β Please upload an audio file.", None
|
|
@@ -42,7 +36,7 @@ def transcribe_audio(audio_file, gemini_key, openrouter_key, model_name):
|
|
| 42 |
duration_min = duration / 60
|
| 43 |
file_size_mb = os.path.getsize(audio_file) / (1024 * 1024)
|
| 44 |
|
| 45 |
-
# Process the audio file
|
| 46 |
output_path, is_zip = process_audio_file(
|
| 47 |
audio_file,
|
| 48 |
gemini_provider,
|
|
@@ -51,7 +45,7 @@ def transcribe_audio(audio_file, gemini_key, openrouter_key, model_name):
|
|
| 51 |
)
|
| 52 |
|
| 53 |
# Determine file type for success message
|
| 54 |
-
if is_zip == "True":
|
| 55 |
file_type = "ZIP archive"
|
| 56 |
file_desc = "Multiple transcript files (chunked audio)"
|
| 57 |
else:
|
|
@@ -73,7 +67,7 @@ def transcribe_audio(audio_file, gemini_key, openrouter_key, model_name):
|
|
| 73 |
|
| 74 |
Click below to download your transcript(s)."""
|
| 75 |
|
| 76 |
-
# Return the file path
|
| 77 |
return success_msg, output_path
|
| 78 |
|
| 79 |
except Exception as e:
|
|
@@ -141,10 +135,10 @@ with gr.Blocks(title="Transcriptinator", theme=gr.themes.Soft()) as app:
|
|
| 141 |
# Status output
|
| 142 |
status_output = gr.Markdown(label="Status")
|
| 143 |
|
| 144 |
-
# Download
|
| 145 |
-
download_output = gr.File(label="π₯ Download Transcript"
|
| 146 |
|
| 147 |
-
# Information section
|
| 148 |
gr.Markdown("""
|
| 149 |
---
|
| 150 |
### π― What you'll get:
|
|
@@ -152,29 +146,6 @@ with gr.Blocks(title="Transcriptinator", theme=gr.themes.Soft()) as app:
|
|
| 152 |
- π **Summary** in 2-3 sentences
|
| 153 |
- π‘ **Key ideas** with descriptions
|
| 154 |
- π **Markdown file** ready to download
|
| 155 |
-
|
| 156 |
-
### π€ AI Models:
|
| 157 |
-
|
| 158 |
-
**Gemini** (Google) - Transcription:
|
| 159 |
-
- Gemini 2.5 Flash (recommended - fastest, best quality)
|
| 160 |
-
- Gemini 2.0 Flash (experimental)
|
| 161 |
-
- Gemini 1.5 Flash (stable)
|
| 162 |
-
- Native audio support with timestamps and speakers
|
| 163 |
-
|
| 164 |
-
**OpenRouter** (Optional) - Summarization:
|
| 165 |
-
- Uses DeepSeek R1 (free, excellent reasoning)
|
| 166 |
-
- Better summaries and key ideas extraction
|
| 167 |
-
- Leave API key empty to use Gemini for everything
|
| 168 |
-
|
| 169 |
-
### π Privacy:
|
| 170 |
-
- Your API keys are never stored
|
| 171 |
-
- Audio files are processed temporarily and deleted
|
| 172 |
-
- All processing happens through your own credentials
|
| 173 |
-
|
| 174 |
-
### π‘ Tips:
|
| 175 |
-
- **New users:** Start with just Gemini API key
|
| 176 |
-
- **Better summaries:** Add OpenRouter key (optional, free)
|
| 177 |
-
- **Large files:** App automatically chunks files >30MB
|
| 178 |
""")
|
| 179 |
|
| 180 |
# Connect the transcription function
|
|
@@ -184,6 +155,7 @@ with gr.Blocks(title="Transcriptinator", theme=gr.themes.Soft()) as app:
|
|
| 184 |
outputs=[status_output, download_output]
|
| 185 |
)
|
| 186 |
|
| 187 |
-
# Launch the app with queuing
|
| 188 |
if __name__ == "__main__":
|
| 189 |
-
|
|
|
|
|
|
| 8 |
from transcribe_core import process_audio_file, get_audio_duration
|
| 9 |
from ai_providers import GeminiProvider, OpenRouterProvider
|
| 10 |
|
| 11 |
+
# Establish absolute paths for Hugging Face Spaces compatibility
|
| 12 |
+
CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
|
| 13 |
+
OUTPUT_FOLDER = os.path.join(CURRENT_DIR, "outputs")
|
| 14 |
|
| 15 |
def transcribe_audio(audio_file, gemini_key, openrouter_key, model_name):
|
| 16 |
"""
|
| 17 |
Main transcription function for Gradio interface.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
"""
|
| 19 |
if not audio_file:
|
| 20 |
return "β Please upload an audio file.", None
|
|
|
|
| 36 |
duration_min = duration / 60
|
| 37 |
file_size_mb = os.path.getsize(audio_file) / (1024 * 1024)
|
| 38 |
|
| 39 |
+
# Process the audio file - ensure this function in core uses absolute paths
|
| 40 |
output_path, is_zip = process_audio_file(
|
| 41 |
audio_file,
|
| 42 |
gemini_provider,
|
|
|
|
| 45 |
)
|
| 46 |
|
| 47 |
# Determine file type for success message
|
| 48 |
+
if str(is_zip) == "True":
|
| 49 |
file_type = "ZIP archive"
|
| 50 |
file_desc = "Multiple transcript files (chunked audio)"
|
| 51 |
else:
|
|
|
|
| 67 |
|
| 68 |
Click below to download your transcript(s)."""
|
| 69 |
|
| 70 |
+
# Return the absolute file path - Gradio handles the download via proxy
|
| 71 |
return success_msg, output_path
|
| 72 |
|
| 73 |
except Exception as e:
|
|
|
|
| 135 |
# Status output
|
| 136 |
status_output = gr.Markdown(label="Status")
|
| 137 |
|
| 138 |
+
# Download component - removed 'interactive=False' for better stability
|
| 139 |
+
download_output = gr.File(label="π₯ Download Transcript")
|
| 140 |
|
| 141 |
+
# Information section ... (remains unchanged)
|
| 142 |
gr.Markdown("""
|
| 143 |
---
|
| 144 |
### π― What you'll get:
|
|
|
|
| 146 |
- π **Summary** in 2-3 sentences
|
| 147 |
- π‘ **Key ideas** with descriptions
|
| 148 |
- π **Markdown file** ready to download
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
""")
|
| 150 |
|
| 151 |
# Connect the transcription function
|
|
|
|
| 155 |
outputs=[status_output, download_output]
|
| 156 |
)
|
| 157 |
|
| 158 |
+
# Launch the app with queuing and allowed_paths for file access
|
| 159 |
if __name__ == "__main__":
|
| 160 |
+
os.makedirs(OUTPUT_FOLDER, exist_ok=True)
|
| 161 |
+
app.queue().launch(allowed_paths=[OUTPUT_FOLDER])
|