debug shenenigans

app.py

@@ -35,8 +35,8 @@ def infer(md, mic_audio, audio, speaker_id, pitch_shift, length_scale, noise_sca
     
     if mic_audio:
         sr, source  = mic_audio
-        source = torch.Tensor(source)
-        source = source.T
+        source = torch.Tensor(source).T
+        print(source.shape)
 
     if audio:
         source, sr = torchaudio.load(audio)
@@ -72,8 +72,8 @@ demo = gradio.Interface(
         
         ( * up to Kiriko and without Bastion. Please forgive. )
         """),
-        gradio.Audio(source="microphone", label="Record your own voice"),
-        gradio.Audio(label="Input Audio", type="filepath"),
+        gradio.Audio(label="Record Input Audio", source="microphone"),
+        gradio.Audio(label="Upload Input Audio", type="filepath"),
         gradio.Dropdown(label="Target Voice", choices=["Ana", "Ashe", "Baptiste", "Brigitte", "Cassidy", "Doomfist", "D.Va", "Echo", "Genji", "Hanzo", "Junker Queen", "Junkrat", "Kiriko", "Lúcio", "Mei", "Mercy", "Moira", "Orisa", "Pharah", "Reaper", "Reinhardt", "Roadhog", "Sigma", "Sojourn", "Soldier_ 76", "Sombra", "Symmetra", "Torbjörn", "Tracer", "Widowmaker", "Winston", "Zarya", "Zenyatta"], type="index", value="Ana"),
         gradio.Slider(label="Pitch Shift Input (+12 = up one octave, ⚠️ broken AF ⚠️)", minimum=-12.0, maximum=12.0, value=0, step=1),
         gradio.Slider(label="Length Factor (higher = slower speech)", minimum=0.1, maximum=2.0, value=1.0),