add audios files

app.py

@@ -61,52 +61,34 @@ def text_to_speech(bambara_text, reference_audio: Optional[Tuple] = None):
 
 
 # Function to enhance speech
-# @spaces.GPU
-# def enhance_speech(audio_array, sampling_rate, solver, nfe, tau, denoise_before_enhancement):
-#     solver = solver.lower()
-#     nfe = int(nfe)
-#     lambd = 0.9 if denoise_before_enhancement else 0.1
-#
-#     @spaces.GPU(duration=360)
-#     def denoise_audio():
-#         try:
-#             return denoise(audio_array, sampling_rate, device)
-#         except Exception as e:
-#             print("> Error while denoising : ", str(e))
-#             return audio_array, sampling_rate
-#
-#     @spaces.GPU(duration=360)
-#     def enhance_audio():
-#         try:
-#             return enhance(audio_array, sampling_rate, device, nfe=nfe, solver=solver, lambd=lambd, tau=tau)
-#         except Exception as e:
-#             print("> Error while enhancement : ", str(e))
-#             return audio_array, sampling_rate
-#
-#     with concurrent.futures.ThreadPoolExecutor() as executor:
-#         future_denoise = executor.submit(denoise_audio)
-#         future_enhance = executor.submit(enhance_audio)
-#
-#         denoised_audio, new_sr1 = future_denoise.result()
-#         enhanced_audio, new_sr2 = future_enhance.result()
-#
-#         # Convert to numpy and return
-#         return (new_sr1, denoised_audio.cpu().numpy()), (new_sr2, enhanced_audio.cpu().numpy())
-
 def enhance_speech(audio_array, sampling_rate, solver, nfe, tau, denoise_before_enhancement):
     solver = solver.lower()
     nfe = int(nfe)
     lambd = 0.9 if denoise_before_enhancement else 0.1
 
-    denoised_audio, new_sr1 = denoise(audio_array, sampling_rate, device)
-    # enhanced_audio, new_sr2 = enhance(audio_array, sampling_rate, device, nfe=nfe, solver=solver, lambd=lambd, tau=tau)
+    def denoise_audio():
+        try:
+            return denoise(audio_array, sampling_rate, device)
+        except Exception as e:
+            print("> Error while denoising : ", str(e))
+            return audio_array, sampling_rate
 
-    # Convert to numpy and return
-    return (
-        (new_sr1, denoised_audio.cpu()),
-        None
-        # (new_sr2, enhanced_audio.cpu().numpy())
-    )
+    def enhance_audio():
+        try:
+            return enhance(audio_array, sampling_rate, device, nfe=nfe, solver=solver, lambd=lambd, tau=tau)
+        except Exception as e:
+            print("> Error while enhancement : ", str(e))
+            return audio_array, sampling_rate
+
+    with concurrent.futures.ThreadPoolExecutor() as executor:
+        future_denoise = executor.submit(denoise_audio)
+        future_enhance = executor.submit(enhance_audio)
+
+        denoised_audio, new_sr1 = future_denoise.result()
+        enhanced_audio, new_sr2 = future_enhance.result()
+
+        # Convert to numpy and return
+        return (new_sr1, denoised_audio.cpu().numpy()), (new_sr2, enhanced_audio.cpu().numpy())
 
 
 def resample_audio(audio_tensor, orig_sr, target_sr):
@@ -157,7 +139,7 @@ def _fn(
         audio_array, sampling_rate = text_to_speech(bambara_text)
 
     # Step 3: Enhance the audio
-    denoised_audio, _ = enhance_speech(
+    denoised_audio, enhanced_audio = enhance_speech(
         audio_array,
         sampling_rate,
         solver,
@@ -166,25 +148,12 @@ def _fn(
         denoise_before_enhancement
     )
 
-    denoised_audio_array = resample_audio(denoised_audio[1], denoised_audio[0], sampling_rate).numpy()
-
-    audio_out = audio_array.numpy()
-    print("audio_array sampling_rate", sampling_rate)
-    print("audio_array type", type(audio_out))
-    print("audio_array shape", audio_out.shape)
-    print("audio_array data", audio_out)
-
-    print("denoised_audio sampling_rate", sampling_rate)
-    print("denoised_audio type", type(denoised_audio_array))
-    print("denoised_audio shape", denoised_audio_array.shape)
-    print("denoised_audio data", denoised_audio_array)
-
     # Return all outputs
     return (
         bambara_text,
         (sampling_rate, audio_array.numpy()),
-        (sampling_rate, denoised_audio_array),
-        # (enhanced_audio[0], convert_to_int16(enhanced_audio[1]))
+        denoised_audio,
+        enhanced_audio
     )