Update app.py

app.py

@@ -102,11 +102,8 @@ async def get_voices_list(proxy=None):
 def find_files(directory):
     file_paths = []
     for filename in os.listdir(directory):
-        # Check if the file has the desired extension
         if filename.endswith('.pth') or filename.endswith('.zip') or filename.endswith('.index'):
-            # If yes, add the file path to the list
             file_paths.append(os.path.join(directory, filename))
-
     return file_paths
 
 
@@ -120,7 +117,6 @@ def unzip_in_folder(my_zip, my_dir):
 
 
 def find_my_model(a_, b_):
-
     if a_ is None or a_.endswith(".pth"):
         return a_, b_
 
@@ -179,7 +175,6 @@ def ensure_valid_file(url):
             raise ValueError("No Content-Length header found")
 
         file_size = int(content_length)
-        # print("debug", url, file_size)
         if file_size > 900000000 and IS_ZERO_GPU:
             raise ValueError("The file is too large. Max allowed is 900 MB.")
 
@@ -196,7 +191,6 @@ def clear_files(directory):
 
 
 def get_my_model(url_data, progress=gr.Progress(track_tqdm=True)):
-
     if not url_data:
         return None, None
 
@@ -250,12 +244,50 @@ def get_my_model(url_data, progress=gr.Progress(track_tqdm=True)):
     except Exception as e:
         raise e
     finally:
-        # time.sleep(10)
-        # shutil.rmtree(directory)
         t = threading.Thread(target=clear_files, args=(directory,))
         t.start()
 
 
+# ========== नया फ़ंक्शन: logs फोल्डर से मॉडल्स की सूची बनाना ==========
+def scan_models():
+    """logs/ फोल्डर के अंदर मौजूद हर सबफोल्डर को एक मॉडल मानकर उसकी .pth और .index फाइल ढूंढें"""
+    logs_dir = "logs"
+    if not os.path.isdir(logs_dir):
+        return []
+    
+    models = []
+    for model_name in os.listdir(logs_dir):
+        model_path = os.path.join(logs_dir, model_name)
+        if not os.path.isdir(model_path):
+            continue
+        # सबफोल्डर में .pth और .index फाइल देखें (नाम फोल्डर जैसा होना चाहिए)
+        pth_file = os.path.join(model_path, f"{model_name}.pth")
+        index_file = os.path.join(model_path, f"{model_name}.index")
+        # अगर exact नाम न मिले तो कोई भी .pth और .index उठा लें (fallback)
+        if not os.path.isfile(pth_file) or not os.path.isfile(index_file):
+            # Fallback: फोल्डर में मौजूद पहली .pth और .index फाइल लें
+            pth_candidates = [f for f in os.listdir(model_path) if f.endswith(".pth")]
+            index_candidates = [f for f in os.listdir(model_path) if f.endswith(".index")]
+            if pth_candidates and index_candidates:
+                pth_file = os.path.join(model_path, pth_candidates[0])
+                index_file = os.path.join(model_path, index_candidates[0])
+            else:
+                continue  # इस फोल्डर में जरूरी फाइलें नहीं हैं
+        models.append((model_name, pth_file, index_file))
+    return models
+
+
+def update_model_selection(model_name):
+    """ड्रॉपडाउन से चुने गए मॉडल के अनुसार pth और index का पथ लौटाएँ"""
+    models = scan_models()
+    for name, pth, idx in models:
+        if name == model_name:
+            return pth, idx
+    return None, None
+
+
+# ========== ऑडियो इफेक्ट और कन्वर्जन फंक्शन (कोई बदलाव नहीं) ==========
+
 def add_audio_effects(audio_list, type_output):
     print("Audio effects")
 
@@ -264,7 +296,6 @@ def add_audio_effects(audio_list, type_output):
         try:
             output_path = f'{os.path.splitext(audio_path)[0]}_effects.{type_output}'
 
-            # Initialize audio effects plugins
             board = Pedalboard(
                 [
                     HighpassFilter(),
@@ -273,7 +304,6 @@ def add_audio_effects(audio_list, type_output):
                  ]
             )
 
-            # Temporary WAV to hold processed data before exporting
             temp_wav = f'{os.path.splitext(audio_path)[0]}_temp.wav'
 
             with AudioFile(audio_path) as f:
@@ -283,11 +313,9 @@ def add_audio_effects(audio_list, type_output):
                         effected = board(chunk, f.samplerate, reset=False)
                         o.write(effected)
 
-            # Convert with pydub to desired output type
             audio_seg = AudioSegment.from_file(temp_wav, format=type_output)
             audio_seg.export(output_path, format=type_output, bitrate=("320k" if type_output == "mp3" else None))
 
-            # Clean up temp file
             os.remove(temp_wav)
 
             result.append(output_path)
@@ -300,7 +328,6 @@ def add_audio_effects(audio_list, type_output):
 
 
 def apply_noisereduce(audio_list, type_output):
-    # https://github.com/sa-if/Audio-Denoiser
     print("Noice reduce")
 
     result = []
@@ -308,16 +335,12 @@ def apply_noisereduce(audio_list, type_output):
         out_path = f"{os.path.splitext(audio_path)[0]}_noisereduce.{type_output}"
 
         try:
-            # Load audio file
             audio = AudioSegment.from_file(audio_path)
 
-            # Convert audio to numpy array
             samples = np.array(audio.get_array_of_samples())
 
-            # Reduce noise
             reduced_noise = nr.reduce_noise(samples, sr=audio.frame_rate, prop_decrease=0.6)
 
-            # Convert reduced noise signal back to audio
             reduced_audio = AudioSegment(
                 reduced_noise.tobytes(), 
                 frame_rate=audio.frame_rate, 
@@ -325,7 +348,6 @@ def apply_noisereduce(audio_list, type_output):
                 channels=audio.channels
             )
 
-            # Save reduced audio to file
             reduced_audio.export(out_path, format=type_output, bitrate=("320k" if type_output == "mp3" else None))
             result.append(out_path)
 
@@ -353,10 +375,10 @@ def convert_now(audio_files, random_tag, converter, type_output, steps):
 
 def run(
     audio_files,
-    file_m,
+    file_m,                # अब यह hidden टेक्स्टबॉक्स से आएगा (pth का पथ)
     pitch_alg,
     pitch_lvl,
-    file_index,
+    file_index,            # hidden टेक्स्टबॉक्स से index का पथ
     index_inf,
     r_m_f,
     e_r,
@@ -378,6 +400,7 @@ def run(
     except Exception as e:
         print(e)
 
+    # अगर file_m टेक्स्ट फाइल है तो पुराने लॉजिक से हैंडल करें
     if file_m is not None and file_m.endswith(".txt"):
         file_m, file_index = find_my_model(file_m, file_index)
         print(file_m, file_index)
@@ -409,6 +432,8 @@ def run(
     return result
 
 
+# ========== UI कॉन्फ़िगरेशन (बदलाव सहित) ==========
+
 def audio_conf():
     return gr.File(
         label="Audio files",
@@ -418,14 +443,26 @@ def audio_conf():
     )
 
 
-def model_conf():
-    return gr.File(
-        label="Model file",
-        type="filepath",
-        height=130,
+def model_dropdown_conf():
+    """मॉडल चुनने के लिए ड्रॉपडाउन बनाएँ"""
+    models = scan_models()
+    choices = [name for name, _, _ in models]
+    return gr.Dropdown(
+        choices=choices,
+        value=choices[0] if choices else None,
+        label="Select Model",
+        interactive=True,
     )
 
 
+# ये दो छुपे हुए टेक्स्टबॉक्स हैं जिनमें असली फाइल पथ जाएँगे
+def hidden_model_path_conf():
+    return gr.Textbox(visible=False)
+
+def hidden_index_path_conf():
+    return gr.Textbox(visible=False)
+
+
 def pitch_algo_conf():
     return gr.Dropdown(
         PITCH_ALGO_OPT,
@@ -448,14 +485,6 @@ def pitch_lvl_conf():
     )
 
 
-def index_conf():
-    return gr.File(
-        label="Index file",
-        type="filepath",
-        height=130,
-    )
-
-
 def index_inf_conf():
     return gr.Slider(
         minimum=0,
@@ -515,7 +544,6 @@ def active_tts_conf():
     return gr.Checkbox(
         False,
         label="TTS",
-        # info="",
         container=False,
     )
 
@@ -551,7 +579,6 @@ def tts_play_conf():
     return gr.Checkbox(
         False,
         label="Play",
-        # info="",
         container=False,
         visible=False,
     )
@@ -561,7 +588,6 @@ def sound_gui():
     return gr.Audio(
         value=None,
         type="filepath",
-        # format="mp3",
         autoplay=True,
         visible=True,
         interactive=False,
@@ -591,7 +617,6 @@ def denoise_conf():
     return gr.Checkbox(
         False,
         label="Denoise",
-        # info="",
         container=False,
         visible=True,
     )
@@ -601,7 +626,6 @@ def effects_conf():
     return gr.Checkbox(
         False,
         label="Reverb",
-        # info="",
         container=False,
         visible=True,
     )
@@ -637,7 +661,6 @@ def down_active_conf():
     return gr.Checkbox(
         False,
         label="URL-to-Model",
-        # info="",
         container=False,
     )
 
@@ -669,9 +692,10 @@ def show_components_down(value_active):
         visible=value_active
     )
 
+
 CSS = """
 #audio_tts {
-  visibility: hidden;   /* invisible but still takes space */
+  visibility: hidden;
   height: 0px;
   width: 0px;
   max-width: 0px;
@@ -684,6 +708,7 @@ def get_gui(theme):
         gr.Markdown(title)
         gr.Markdown(description)
 
+        # ===== TTS सेक्शन (कोई बदलाव नहीं) =====
         active_tts = active_tts_conf()
         with gr.Row():
             with gr.Column(scale=1):
@@ -705,7 +730,6 @@ def get_gui(theme):
         )
 
         aud = audio_conf()
-        # gr.HTML("<hr>")
 
         tts_button.click(
             fn=infer_tts_audio,
@@ -713,144 +737,7 @@ def get_gui(theme):
             outputs=[aud, tts_play],
         )
 
+        # ===== URL-to-Model सेक्शन (कोई बदलाव नहीं) =====
         down_active_gui = down_active_conf()
         down_info = gr.Markdown(
-            f"Provide a link to a zip file, like this one: `https://huggingface.co/MrDawg/ToothBrushing/resolve/main/ToothBrushing.zip?download=true`, or separate links with a comma for the .pth and .index files, like this: `{test_model}`",
-            visible=False
-        )
-        with gr.Row():
-            with gr.Column(scale=3):
-                down_url_gui = down_url_conf()
-            with gr.Column(scale=1):
-                down_button_gui = down_button_conf()
-
-        with gr.Column():
-            with gr.Row():
-                model = model_conf()
-                indx = index_conf()
-
-        down_active_gui.change(
-            show_components_down,
-            [down_active_gui],
-            [down_info, down_url_gui, down_button_gui]
-        )
-
-        down_button_gui.click(
-            get_my_model,
-            [down_url_gui],
-            [model, indx]
-        )
-
-        with gr.Accordion(label="Advanced settings", open=False):
-            algo = pitch_algo_conf()
-            algo_lvl = pitch_lvl_conf()
-            indx_inf = index_inf_conf()
-            res_fc = respiration_filter_conf()
-            envel_r = envelope_ratio_conf()
-            const = consonant_protec_conf()
-            steps_gui = steps_conf()
-            format_out = format_output_gui()
-            with gr.Row():
-                with gr.Column():
-                    with gr.Row():
-                        denoise_gui = denoise_conf()
-                        effects_gui = effects_conf()
-        button_base = button_conf()
-        output_base = output_conf()
-
-        button_base.click(
-            run,
-            inputs=[
-                aud,
-                model,
-                algo,
-                algo_lvl,
-                indx,
-                indx_inf,
-                res_fc,
-                envel_r,
-                const,
-                denoise_gui,
-                effects_gui,
-                format_out,
-                steps_gui,
-            ],
-            outputs=[output_base],
-        )
-
-        gr.Examples(
-            examples=[
-                [
-                    ["./test.ogg"],
-                    "./model.pth",
-                    "rmvpe+",
-                    0,
-                    "./model.index",
-                    0.75,
-                    3,
-                    0.25,
-                    0.50,
-                ],
-                [
-                    ["./example2/test2.ogg"],
-                    "./example2/model_link.txt",
-                    "rmvpe+",
-                    0,
-                    "./example2/index_link.txt",
-                    0.75,
-                    3,
-                    0.25,
-                    0.50,
-                ],
-                [
-                    ["./example3/test3.wav"],
-                    "./example3/zip_link.txt",
-                    "rmvpe+",
-                    0,
-                    None,
-                    0.75,
-                    3,
-                    0.25,
-                    0.50,
-                ],
-
-            ],
-            fn=run,
-            inputs=[
-                aud,
-                model,
-                algo,
-                algo_lvl,
-                indx,
-                indx_inf,
-                res_fc,
-                envel_r,
-                const,
-            ],
-            outputs=[output_base],
-            cache_examples=False,
-        )
-        gr.Markdown(RESOURCES)
-
-    return app
-
-
-if __name__ == "__main__":
-    tts_voice_list = asyncio.new_event_loop().run_until_complete(get_voices_list(proxy=None))
-    voices = sorted([
-        (" - ".join(reversed(v["FriendlyName"].split("-"))).replace("Microsoft ", "").replace("Online (Natural)", f"({v['Gender']})").strip(), f"{v['ShortName']}-{v['Gender']}")
-        for v in tts_voice_list
-    ])
-
-    app = get_gui(theme)
-
-    app.queue(default_concurrency_limit=40)
-
-    app.launch(
-        max_threads=40,
-        share=IS_COLAB,
-        show_error=True,
-        quiet=False,
-        debug=IS_COLAB,
-        ssr_mode=False,
-    )
+            f"Provide a link to a zip file, like this one: `https://huggingface.co/MrDawg/ToothBrushing/resolve/main/ToothBrushing.zip?download=true`, or