Create app.py

app.py

@@ -0,0 +1,49 @@
+import gradio as gr
+import numpy as np
+import io
+import os
+from openai import OpenAI
+from pydub import AudioSegment
+from pydub.playback import play
+
+
+# Set an environment variable for key
+os.environ['OPENAI_API_KEY'] = os.environ.get('OPENAI_API_KEY')
+
+client = OpenAI() # add api_key
+
+def stream_and_yield_audio(text, model, voice):
+    response = client.audio.speech.create(
+        model=model, #"tts-1", for example
+        voice=voice , #"alloy", for example
+        input=text,
+    )
+
+    # Convert the binary response content to a byte stream
+    byte_stream = io.BytesIO(response.content)
+
+    # Read the audio data from the byte stream
+    audio = AudioSegment.from_file(byte_stream, format="mp3")
+
+    # Export the audio as WAV format
+    sample_width = audio.sample_width
+    sample_rate = audio.frame_rate
+    audio_data = np.array(audio.get_array_of_samples(), dtype=np.int16)
+
+    # Yield the audio data
+    yield sample_rate, audio_data #audio_data.tobytes(), sample_width
+    
+
+# demo using older gradio version (3.50.2) 
+with gr.Blocks() as demo:
+    with gr.Row():
+      model = gr.Dropdown(choices=['tts-1','tts-1-hd'], label='Model', value='tts-1')
+      voice = gr.Dropdown(choices=['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'], label='Voice Options', value='alloy')
+
+    text = gr.Textbox(label="Input text")
+    btn = gr.Button("Greet")
+    output_audio = gr.Audio(label="Speech Output", streaming=True, autoplay=True)
+    
+    btn.click(fn=stream_and_yield_audio, inputs=[text,model, voice], outputs=output_audio, api_name="tts-stream")
+
+demo.queue().launch()