Upload 2 files

app.py

@@ -0,0 +1,31 @@
+import gradio as gr
+import soundfile as sf
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Tokenizer
+
+# Load the pre-trained model and tokenizer
+model_name = "facebook/wav2vec2-large-960h-lv60-self"
+tokenizer = Wav2Vec2Tokenizer.from_pretrained(model_name)
+model = SpeechRecognitionModel("jonatasgrosman/wav2vec2-large-xlsr-53-english")
+
+# Define the speech recognition function
+def transcribe_audio(audio):
+    audio_path = "audio.wav"
+    sf.write(audio_path, audio, samplerate=16000)
+    transcriptions = model.transcribe(audio_path)
+    return transcriptions["transcription"]
+
+# Set up the Gradio interface
+audio_input = gr.inputs.Audio(source="microphone", type="numpy")
+text_output = gr.outputs.Textbox()
+
+interface = gr.Interface(
+    fn=transcribe_audio,
+    inputs=audio_input,
+    outputs=text_output,
+    title="Speech Recognition",
+    description="Transcribe speech in real-time.",
+    server_port=8000,
+)
+
+if __name__ == "__main__":
+    interface.launch()

requirements.txt

@@ -0,0 +1,6 @@
+gradio==2.3.0
+torch
+transformers
+torchaudio
+huggingsound
+soundfile