Upload app.py

app.py

@@ -1,173 +1,325 @@
 import gradio as gr
 import whisper
+import os
 
-model = None
+models = {}
 
-def load_model():
-    global model
-    if model is None:
-        model = whisper.load_model("base")
-    return model
+def load_model(model_name):
+    global models
+    if model_name not in models:
+        models[model_name] = whisper.load_model(model_name)
+    return models[model_name]
 
-def transcribe(file):
+def transcribe(file, model_name, language, show_timestamps, task):
     if file is None:
-        return "Please upload a video or audio file."
+        return "Please upload a video or audio file.", ""
+
     try:
-        m = load_model()
-        result = m.transcribe(file.name, language="en", verbose=False)
-        return result["text"].strip()
+        m = load_model(model_name)
+
+        lang = None if language == "Auto Detect" else language.lower()
+
+        result = m.transcribe(
+            file.name,
+            language=lang,
+            verbose=False,
+            task="translate" if task == "Translate to English" else "transcribe"
+        )
+
+        # Full plain transcript
+        plain = result["text"].strip()
+
+        # Timestamped transcript
+        if show_timestamps:
+            lines = []
+            for seg in result["segments"]:
+                start = format_time(seg["start"])
+                end = format_time(seg["end"])
+                lines.append(f"[{start} → {end}]  {seg['text'].strip()}")
+            timestamped = "\n".join(lines)
+        else:
+            timestamped = plain
+
+        return timestamped, plain
+
     except Exception as e:
-        return f"Error: {str(e)}"
+        return f"Error: {str(e)}", ""
+
+def format_time(seconds):
+    m = int(seconds // 60)
+    s = int(seconds % 60)
+    ms = int((seconds % 1) * 10)
+    return f"{m:02d}:{s:02d}.{ms}"
+
+def save_transcript(text):
+    if not text:
+        return None
+    path = "/tmp/transcript.txt"
+    with open(path, "w", encoding="utf-8") as f:
+        f.write(text)
+    return path
 
 custom_css = """
-@import url('https://fonts.googleapis.com/css2?family=DM+Serif+Display:ital@0;1&family=DM+Sans:wght@300;400;500&display=swap');
+@import url('https://fonts.googleapis.com/css2?family=DM+Serif+Display:ital@0;1&family=DM+Sans:opsz,wght@9..40,300;9..40,400;9..40,500&display=swap');
 
 * { box-sizing: border-box; }
 
 body, .gradio-container {
-    background: #0a0a0a !important;
+    background: #080808 !important;
     font-family: 'DM Sans', sans-serif !important;
 }
 
 .gradio-container {
-    max-width: 920px !important;
+    max-width: 1000px !important;
     margin: 0 auto !important;
-    padding: 48px 24px !important;
+    padding: 48px 28px 80px !important;
 }
 
+/* Gold bar */
+.gradio-container::before {
+    content: '';
+    display: block;
+    width: 36px;
+    height: 2px;
+    background: #c9a96e;
+    margin-bottom: 32px;
+}
+
+/* Title */
 .prose h1 {
     font-family: 'DM Serif Display', serif !important;
-    font-size: 2.8rem !important;
+    font-size: 3rem !important;
     font-weight: 400 !important;
     color: #f0ece2 !important;
     letter-spacing: -0.02em !important;
-    line-height: 1.1 !important;
+    line-height: 1.05 !important;
     margin-bottom: 0 !important;
 }
 
 .prose p {
-    font-family: 'DM Sans', sans-serif !important;
-    color: #6b6860 !important;
-    font-size: 0.95rem !important;
+    color: #555 !important;
+    font-size: 0.9rem !important;
     font-weight: 300 !important;
-    margin-top: 10px !important;
+    margin-top: 8px !important;
+    letter-spacing: 0.01em !important;
 }
 
-.contain, .gap {
-    background: transparent !important;
-    border: none !important;
-    gap: 20px !important;
-}
+/* Panels */
+.contain, .gap { background: transparent !important; border: none !important; }
 
 .block {
-    background: #111111 !important;
-    border: 1px solid #1e1e1e !important;
-    border-radius: 16px !important;
+    background: #0f0f0f !important;
+    border: 1px solid #1c1c1c !important;
+    border-radius: 14px !important;
+    transition: border-color 0.2s !important;
 }
 
-.block label span {
+.block:hover { border-color: #2a2a2a !important; }
+
+/* Labels */
+.block label > span, label > span {
     font-family: 'DM Sans', sans-serif !important;
-    font-size: 0.75rem !important;
+    font-size: 0.72rem !important;
     font-weight: 500 !important;
-    color: #6b6860 !important;
+    color: #4a4a4a !important;
     text-transform: uppercase !important;
-    letter-spacing: 0.1em !important;
+    letter-spacing: 0.12em !important;
+}
+
+/* Dropdowns & selects */
+select, .wrap-inner, input[type="text"] {
+    background: #0f0f0f !important;
+    color: #c8c4bc !important;
+    border: 1px solid #1c1c1c !important;
+    border-radius: 10px !important;
+    font-family: 'DM Sans', sans-serif !important;
+    font-size: 0.88rem !important;
 }
 
+/* Textarea */
 textarea {
     background: transparent !important;
-    color: #f0ece2 !important;
+    color: #d4d0c8 !important;
     font-family: 'DM Sans', sans-serif !important;
-    font-size: 0.95rem !important;
-    line-height: 1.8 !important;
+    font-size: 0.92rem !important;
+    line-height: 1.85 !important;
     font-weight: 300 !important;
     border: none !important;
+    letter-spacing: 0.01em !important;
 }
 
-.file-preview, .upload-container {
-    background: transparent !important;
-    border: none !important;
-    color: #555 !important;
-}
+textarea::placeholder { color: #2e2e2e !important; }
 
+/* Checkbox */
+input[type="checkbox"] { accent-color: #c9a96e !important; }
+
+/* Primary button */
 button.primary {
     background: #c9a96e !important;
     border: none !important;
-    border-radius: 12px !important;
-    color: #0a0a0a !important;
+    border-radius: 11px !important;
+    color: #080808 !important;
     font-family: 'DM Sans', sans-serif !important;
-    font-size: 0.8rem !important;
+    font-size: 0.78rem !important;
     font-weight: 500 !important;
-    letter-spacing: 0.1em !important;
+    letter-spacing: 0.12em !important;
     text-transform: uppercase !important;
     padding: 14px 28px !important;
     transition: all 0.2s ease !important;
+    box-shadow: 0 2px 12px rgba(201,169,110,0.15) !important;
 }
 
 button.primary:hover {
     background: #d4b87e !important;
     transform: translateY(-1px) !important;
-    box-shadow: 0 4px 20px rgba(201,169,110,0.25) !important;
+    box-shadow: 0 6px 24px rgba(201,169,110,0.3) !important;
 }
 
+button.primary:active { transform: translateY(0) !important; }
+
+/* Secondary button */
 button.secondary {
     background: transparent !important;
-    border: 1px solid #222 !important;
-    border-radius: 12px !important;
-    color: #555 !important;
+    border: 1px solid #1e1e1e !important;
+    border-radius: 11px !important;
+    color: #4a4a4a !important;
     font-family: 'DM Sans', sans-serif !important;
-    font-size: 0.8rem !important;
+    font-size: 0.78rem !important;
     font-weight: 500 !important;
-    letter-spacing: 0.1em !important;
+    letter-spacing: 0.12em !important;
     text-transform: uppercase !important;
     transition: all 0.2s ease !important;
 }
 
 button.secondary:hover {
-    border-color: #444 !important;
-    color: #f0ece2 !important;
+    border-color: #3a3a3a !important;
+    color: #c8c4bc !important;
 }
 
-footer { display: none !important; }
+/* Download button area */
+.file-preview {
+    background: transparent !important;
+    border: none !important;
+}
 
-textarea::-webkit-scrollbar { width: 4px; }
-textarea::-webkit-scrollbar-track { background: transparent; }
-textarea::-webkit-scrollbar-thumb { background: #2a2a2a; border-radius: 4px; }
+/* Tabs */
+.tab-nav button {
+    font-family: 'DM Sans', sans-serif !important;
+    font-size: 0.78rem !important;
+    font-weight: 500 !important;
+    letter-spacing: 0.1em !important;
+    text-transform: uppercase !important;
+    color: #4a4a4a !important;
+    background: transparent !important;
+    border: none !important;
+    border-bottom: 2px solid transparent !important;
+    padding: 10px 16px !important;
+    transition: all 0.2s !important;
+}
 
-.gradio-container::before {
-    content: '';
-    display: block;
-    width: 36px;
-    height: 2px;
-    background: #c9a96e;
-    margin-bottom: 28px;
+.tab-nav button.selected {
+    color: #c9a96e !important;
+    border-bottom-color: #c9a96e !important;
 }
+
+/* Divider */
+.divider { border-color: #1c1c1c !important; }
+
+/* Progress */
+.progress-bar { background: #c9a96e !important; }
+.progress-bar-wrap { background: #1c1c1c !important; }
+
+/* Scrollbar */
+::-webkit-scrollbar { width: 3px; }
+::-webkit-scrollbar-track { background: transparent; }
+::-webkit-scrollbar-thumb { background: #2a2a2a; border-radius: 4px; }
+
+footer { display: none !important; }
+
+/* Settings row */
+.settings-row { gap: 12px !important; }
 """
 
+LANGUAGES = [
+    "Auto Detect", "English", "Hindi", "Spanish", "French",
+    "German", "Italian", "Portuguese", "Chinese", "Japanese",
+    "Korean", "Arabic", "Russian", "Dutch", "Turkish"
+]
+
 with gr.Blocks(css=custom_css, title="KalpTranscript") as demo:
+
     gr.Markdown("""
 # KalpTranscript
 Turn any video or audio into clean, readable text — powered by OpenAI Whisper.
 """)
 
     with gr.Row():
-        with gr.Column():
+        # Left column - controls
+        with gr.Column(scale=1):
             file_input = gr.File(
                 label="Drop your file here",
                 file_types=["video", "audio"]
             )
+
+            with gr.Row(elem_classes=["settings-row"]):
+                model_choice = gr.Dropdown(
+                    choices=["tiny", "base", "small", "medium"],
+                    value="small",
+                    label="Model",
+                    info="small = fast & accurate, medium = best quality"
+                )
+                language = gr.Dropdown(
+                    choices=LANGUAGES,
+                    value="Auto Detect",
+                    label="Language"
+                )
+
+            with gr.Row():
+                task = gr.Radio(
+                    choices=["Transcribe", "Translate to English"],
+                    value="Transcribe",
+                    label="Task"
+                )
+
+            timestamps = gr.Checkbox(
+                label="Show timestamps",
+                value=False
+            )
+
             with gr.Row():
                 clear_btn = gr.ClearButton(value="Clear")
                 submit_btn = gr.Button("Transcribe →", variant="primary")
 
-        with gr.Column():
-            output = gr.Textbox(
-                label="Transcript",
-                lines=14,
-                placeholder="Your transcript will appear here..."
-            )
+        # Right column - output
+        with gr.Column(scale=1):
+            with gr.Tabs():
+                with gr.Tab("Transcript"):
+                    output = gr.Textbox(
+                        label="",
+                        lines=16,
+                        placeholder="Your transcript will appear here...",
+                        show_copy_button=False
+                    )
+                with gr.Tab("Download"):
+                    plain_output = gr.Textbox(
+                        label="Plain text (for download)",
+                        lines=10,
+                        visible=False
+                    )
+                    download_btn = gr.Button("Save as .txt", variant="secondary")
+                    download_file = gr.File(label="Download", visible=True)
+
+    submit_btn.click(
+        fn=transcribe,
+        inputs=[file_input, model_choice, language, timestamps, task],
+        outputs=[output, plain_output]
+    )
 
-    submit_btn.click(fn=transcribe, inputs=file_input, outputs=output)
+    download_btn.click(
+        fn=save_transcript,
+        inputs=plain_output,
+        outputs=download_file
+    )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()