Rename app.txt to app.py

app.py

@@ -0,0 +1,19 @@
+import tempfile ,os
+import gradio as gr
+from transformers import VitsModel, AutoTokenizer,pipeline
+import torch
+import numpy as np
+import torchaudio
+
+# Load model directly
+from transformers import AutoProcessor, AutoModelForCTC
+
+processor = AutoProcessor.from_pretrained("SeyedAli/Persian-Speech-Transcription-Wav2Vec2-V1")
+model = AutoModelForCTC.from_pretrained("SeyedAli/Persian-Speech-Transcription-Wav2Vec2-V1")
+
+def ASR(audio):
+   pipe = pipeline("automatic-speech-recognition", model="SeyedAli/Persian-Speech-Transcription-Wav2Vec2-V1")
+   text=pipe(torchaudio.load(audio))
+   return text
+iface = gr.Interface(fn=TTS, inputs="audio", outputs="text")
+iface.launch(share=False)

app.txt

@@ -1,22 +0,0 @@
-import tempfile ,os
-import gradio as gr
-from transformers import VitsModel, AutoTokenizer,pipeline
-import torch
-import numpy as np
-import torchaudio
-
-model = VitsModel.from_pretrained("SeyedAli/Persian-Speech-synthesis")
-tokenizer = AutoTokenizer.from_pretrained("SeyedAli/Persian-Speech-synthesis")
-
-def ASR(audio):
-   inputs = tokenizer(text, return_tensors="pt")
-   with torch.no_grad():
-        output = model(**inputs).waveform 
-   with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
-        torchaudio.save(fp, output, model.config.sampling_rate,format="wav")
-        return fp.name
-iface = gr.Interface(fn=TTS, inputs="text", outputs="audio")
-iface.launch(share=False)
-pipe = pipeline("automatic-speech-recognition", model=Model)
-    prediction = asr(audio, chunk_length_s=5, stride_length_s=1)
-    return prediction["text"]