Update app.py

app.py

@@ -3,9 +3,7 @@ from transformers import pipeline
 from transformers.pipelines.audio_utils import ffmpeg_read
 import gradio as gr
 
-# Luego, utiliza gi.Audio en lugar de gr.inputs.Audio
-
-MODEL_NAME = "ovieyra21/whisper-small-curso"
+MODEL_NAME = "openai/whisper-small"
 BATCH_SIZE = 8
 
 device = 0 if torch.cuda.is_available() else "cpu"
@@ -17,6 +15,7 @@ pipe = pipeline(
     device=device,
 )
 
+
 # Copied from https://github.com/openai/whisper/blob/c09a7ae299c4c34c5839a76380ae407e7d785914/whisper/utils.py#L50
 def format_timestamp(seconds: float, always_include_hours: bool = False, decimal_marker: str = "."):
     if seconds is not None:
@@ -37,6 +36,7 @@ def format_timestamp(seconds: float, always_include_hours: bool = False, decimal
         # we have a malformed timestamp so just return it as is
         return seconds
 
+
 def transcribe(file, task, return_timestamps):
     outputs = pipe(file, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=return_timestamps)
     text = outputs["text"]
@@ -70,10 +70,11 @@ mic_transcribe = gr.Interface(
     ),
     allow_flagging="never",
 )
+
 file_transcribe = gr.Interface(
     fn=transcribe,
     inputs=[
-        gr.inputs.Audio(source="upload", optional=True, label="Audio file", type="audio"),
+        gr.inputs.Audio(source="upload", optional=True, label="Audio file", type="filepath"),
         gr.inputs.Radio(["transcribe", "translate"], label="Task", default="transcribe"),
         gr.inputs.Checkbox(default=False, label="Return timestamps"),
     ],
@@ -97,6 +98,4 @@ file_transcribe = gr.Interface(
 with demo:
     gr.TabbedInterface([mic_transcribe, file_transcribe], ["Transcribe Microphone", "Transcribe Audio File"])
 
-
-
 demo.launch(enable_queue=True)