Update app.py

app.py

@@ -130,13 +130,16 @@ def return_prediction_compare(mic=None, file=None, device=device):
 
 # Create a partial function with the device pre-applied
 return_prediction_whisper_with_device = partial(return_prediction_whisper, device=device)
-
+return_prediction_w2v2_with_device = partial(return_prediction_w2v2, device=device)
 
 # Load the ASR models
 whisper_classifier = foreign_class(source="Macedonian-ASR/whisper-large-v3-macedonian-asr", pymodule_file="custom_interface_app.py", classname="ASR")
 whisper_classifier = whisper_classifier.to(device)
 whisper_classifier.eval()
 
+w2v2_classifier = foreign_class(source="Macedonian-ASR/wav2vec2-aed-macedonian-asr", pymodule_file="custom_interface_app.py", classname="ASR")
+w2v2_classifier = w2v2_classifier.to(device)
+w2v2_classifier.eval()
 
 # Load the T5 tokenizer and model for restoring capitalization
 recap_model_name = "Macedonian-ASR/mt5-restore-capitalization-macedonian"
@@ -162,6 +165,13 @@ mic_transcribe_whisper = gr.Interface(
 #     live=False
 # )
 
+mic_transcribe_w2v2 = gr.Interface(
+    fn=return_prediction_w2v2_with_device,
+    inputs=gr.Audio(sources="microphone", type="filepath"),
+    outputs=gr.Textbox(),
+    allow_flagging="never",
+    live=False,
+)
 
 project_description = '''
 <img src="https://i.imghippo.com/files/JXadQ1728417387.png"
@@ -214,8 +224,8 @@ with transcriber_app:
     # state = gr.State(value=[], delete_callback=lambda v: print("STATE DELETED"))
 
     gr.TabbedInterface(
-        [mic_transcribe_whisper],
-        ["Буки-Whisper транскрипција"],
+        [mic_transcribe_whisper, mic_transcribe_w2v2],
+        ["Буки-Whisper транскрипција", "Буки-Wav2vec2 транскрипција"],
     )
     state = gr.State(value=[], delete_callback=lambda v: print("STATE DELETED"))