Update app.py

app.py

@@ -1,50 +1,83 @@
-import gradio as gr
-import os
-import tempfile
-from openai import OpenAI
+import random 
+import gradio as gr 
+import numpy as np 
+from elevenlabs import voices, generate, set_api_key, UnauthenticatedRateLimitError
 
+def pad_buffer(audio):
+    # Pad buffer to multiple of 2 bytes
+    buffer_size = len(audio)
+    element_size = np.dtype(np.int16).itemsize
+    if buffer_size % element_size != 0:
+        audio = audio + b'\0' * (element_size - (buffer_size % element_size))
+    return audio 
 
-def tts(text, model, voice, api_key):
-    if api_key == '':
-        raise gr.Error('Please enter your OpenAI API Key')
-    else:
-        try:
-            client = OpenAI(api_key=api_key)
-
-            response = client.audio.speech.create(
-                model=model, # "tts-1","tts-1-hd"
-                voice=voice, # 'alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'
-                input=text,
-            )
-
-        except Exception as error:
-            # Handle any exception that occurs
-            raise gr.Error("An error occurred while generating speech. Please check your API key and try again.")
-            print(str(error))
-
-    # Create a temp file to save the audio
-    with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as temp_file:
-        temp_file.write(response.content)
+def generate_voice(text, voice_name, api_key):
+    set_api_key(api_key) #set API key
+    try:
+        audio = generate(
+            text[:4000], # Limit to 4000 characters
+            voice=voice_name, 
+            model="eleven_multilingual_v2"
+        )
+        return (44100, np.frombuffer(pad_buffer(audio), dtype=np.int16))
+    except UnauthenticatedRateLimitError as e:
+        raise gr.Error("Thanks for trying out ElevenLabs TTS! You've reached the free tier limit. Please provide an API key to continue.") 
+    except Exception as e:
+        raise gr.Error(e)
+    
+description = """
+Eleven Multilingual V2 is the world's best Text-to-Speech model. Features 38 voices and supports 28 languages. Sign up on [ElevenLabs](https://elevenlabs.io/?from=partnerpierce7156) to get an API Key.
+"""
 
-    # Get the file path of the temp file
-    temp_file_path = temp_file.name
+with gr.Blocks() as block:
+    #gr.Markdown('[ ![ElevenLabs](https://user-images.githubusercontent.com/12028621/262629275-4f85c9cf-85b6-435e-ab50-5b8c7c4e9dd2.png) ](https://elevenlabs.io)')
+    gr.Markdown("# <center> ElevenLabs </center>")
+    gr.Markdown(description)
+        
 
-    return temp_file_path
+    with gr.Row(variant='panel'):
+        input_api_key = gr.Textbox(
+        type='password', 
+        label='Elevenlabs API Key', 
+        placeholder='Enter your API key',
+        elem_id="input_api_key"
+        )
 
+        all_voices = voices() 
+        input_voice = gr.Dropdown(
+            [ voice.name for voice in all_voices ], 
+            value="Rachel",
+            label="Voice", 
+            elem_id="input_voice"
+        )
 
-with gr.Blocks() as demo:
-    gr.Markdown("# <center> OpenAI </center>")
-    #gr.HTML("You can also access the Streaming demo for OpenAI TTS by clicking this <a href='https://huggingface.co/spaces/ysharma/OpenAI_TTS_Streaming'>Gradio demo link</a>")
-    with gr.Row(variant='panel'):
-      api_key = gr.Textbox(type='password', label='OpenAI API Key', placeholder='Enter your API key')
-      model = gr.Dropdown(choices=['tts-1','tts-1-hd'], label='Model', value='tts-1-hd')
-      voice = gr.Dropdown(choices=['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'], label='Voice Options', value='nova')
+    input_text = gr.Textbox(
+        label="Input Text (4000 characters max)", 
+        lines=1, 
+        value="Hello! 你好! Hola! नमस्ते! Bonjour! こんにちは! مرحبا! 안녕하세요! Ciao! Cześć! Привіт! Γειά σας! Здравей! வணக்கம்!",
+        elem_id="input_text"
+    )
+    
+    run_button = gr.Button(
+        text="Generate Voice", 
+        type="button"
+    )
 
-    text = gr.Textbox(label="Input text", placeholder="Enter your text and then click on the button below.")
-    btn = gr.Button("Text-To-Speech")
-    output_audio = gr.Audio(label="Speech Output")
+    out_audio = gr.Audio(
+        label="Generated Voice",
+        type="numpy", 
+        elem_id="out_audio",
+        format="mp3"
+    )
+        
+    inputs = [input_text, input_voice, input_api_key]
+    outputs = [out_audio]
     
-    text.submit(fn=tts, inputs=[text, model, voice, api_key], outputs=output_audio, api_name="tts_enter_key", concurrency_limit=None)
-    btn.click(fn=tts, inputs=[text, model, voice, api_key], outputs=output_audio, api_name="tts_button", concurrency_limit=None)
+    run_button.click(
+        fn=generate_voice, 
+        inputs=inputs, 
+        outputs=outputs, 
+        queue=True
+    )
 
-demo.launch()
+block.queue(concurrency_count=5).launch(debug=True)