add audios files

app.py

@@ -96,6 +96,15 @@ def enhance_speech(audio_array, sampling_rate, solver, nfe, tau, denoise_before_
         return (new_sr1, denoised_audio.cpu().numpy()), (new_sr2, enhanced_audio.cpu().numpy())
 
 
+def convert_to_int16(audio_array):
+    if audio_array.dtype == torch.float32:
+        # Assuming audio_array values are in the range [-1.0, 1.0]
+        # Scale to int16 range and convert the datatype
+        audio_array = (audio_array * 32767).to(torch.int16)
+
+    return audio_array
+
+
 # Define the Gradio interface
 def _fn(
         src_lang,
@@ -130,11 +139,16 @@ def _fn(
     print("Audio Array Shape:", audio_array.shape)
     print("Sample Rate:", sampling_rate)
     print("Audio Array Dtype:", audio_array.dtype)
-    print("Max Value in Audio Array:", np.max(audio_array))
-    print("Min Value in Audio Array:", np.min(audio_array))
+    print("Max Value in Audio Array:", torch.max(audio_array))
+    print("Min Value in Audio Array:", torch.min(audio_array))
 
     # Return all outputs
-    return bambara_text, (sampling_rate, audio_array.numpy()), denoised_audio, enhanced_audio
+    return (
+        bambara_text,
+        (sampling_rate, convert_to_int16(audio_array).numpy()),
+        (denoised_audio[0], convert_to_int16(denoised_audio[1])),
+        (enhanced_audio[0], convert_to_int16(enhanced_audio[1]))
+    )
 
 
 def main():