Update app.py
Browse files
app.py
CHANGED
|
@@ -1,46 +1,45 @@
|
|
| 1 |
-
|
| 2 |
-
import
|
| 3 |
-
import
|
| 4 |
-
import
|
| 5 |
-
from openai import OpenAI
|
| 6 |
|
| 7 |
# Initialize OpenAI API client with API key
|
| 8 |
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
|
| 9 |
|
| 10 |
# Define the function that generates the speech
|
| 11 |
def generate_speech(input_text):
|
| 12 |
-
# Build completion with OpenAI
|
| 13 |
completion = client.chat.completions.create(
|
| 14 |
-
model="gpt-3.5-turbo-1106",
|
| 15 |
-
messages=[
|
| 16 |
-
{
|
| 17 |
-
|
| 18 |
-
|
|
|
|
|
|
|
| 19 |
]
|
| 20 |
)
|
| 21 |
|
| 22 |
# Extract generated text (response by the assistant) from OpenAI's API response
|
| 23 |
message_content = completion.choices[0].message.content.strip() # Remove whitespace from the ends
|
| 24 |
|
| 25 |
-
#
|
| 26 |
-
#text_to_speech = message_content
|
| 27 |
-
|
| 28 |
response = client.audio.speech.create(
|
| 29 |
-
model="tts-1",
|
| 30 |
-
voice="alloy",
|
| 31 |
-
input=str(message_content)
|
| 32 |
-
|
| 33 |
|
| 34 |
-
|
| 35 |
return response.content
|
| 36 |
-
|
| 37 |
-
# Define the Gradio interface
|
| 38 |
iface = gr.Interface(
|
| 39 |
-
fn=generate_speech,
|
| 40 |
-
inputs=gr.Textbox(),
|
| 41 |
-
outputs=gr.Audio(autoplay=True), #
|
| 42 |
-
live=False
|
| 43 |
)
|
| 44 |
|
| 45 |
# Launch the interface
|
| 46 |
-
iface.launch(show_api=False)
|
|
|
|
| 1 |
+
# Import necessary libraries
|
| 2 |
+
import gradio as gr # Gradio for creating simple UI for functions
|
| 3 |
+
import os # For accessing environment variables
|
| 4 |
+
from openai import OpenAI # OpenAI API client
|
|
|
|
| 5 |
|
| 6 |
# Initialize OpenAI API client with API key
|
| 7 |
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
|
| 8 |
|
| 9 |
# Define the function that generates the speech
|
| 10 |
def generate_speech(input_text):
|
| 11 |
+
# Build conversation completion with the OpenAI chat model
|
| 12 |
completion = client.chat.completions.create(
|
| 13 |
+
model="gpt-3.5-turbo-1106", # Specify the model to use
|
| 14 |
+
messages=[ # This is the conversation history with user input appended
|
| 15 |
+
{
|
| 16 |
+
"role": "system", # System message with instructions for the AI
|
| 17 |
+
"content": (os.getenv("PROMPT"))
|
| 18 |
+
},
|
| 19 |
+
{"role": "user", "content": input_text} # The user's input message
|
| 20 |
]
|
| 21 |
)
|
| 22 |
|
| 23 |
# Extract generated text (response by the assistant) from OpenAI's API response
|
| 24 |
message_content = completion.choices[0].message.content.strip() # Remove whitespace from the ends
|
| 25 |
|
| 26 |
+
# Use OpenAI's text-to-speech API to convert the text response to audio
|
|
|
|
|
|
|
| 27 |
response = client.audio.speech.create(
|
| 28 |
+
model="tts-1", # Specify the TTS model to use
|
| 29 |
+
voice="alloy", # Specify the voice model to use
|
| 30 |
+
input=str(message_content) # The text we want to convert to speech
|
| 31 |
+
)
|
| 32 |
|
| 33 |
+
# Return the binary audio data received from OpenAI
|
| 34 |
return response.content
|
| 35 |
+
|
| 36 |
+
# Define the Gradio interface with our function
|
| 37 |
iface = gr.Interface(
|
| 38 |
+
fn=generate_speech, # Function to invoke
|
| 39 |
+
inputs=gr.Textbox(), # Input element in the UI: a textbox for user text
|
| 40 |
+
outputs=gr.Audio(autoplay=True), # Output element in the UI: an audio player
|
| 41 |
+
live=False # Whether or not to automatically call the function on input change
|
| 42 |
)
|
| 43 |
|
| 44 |
# Launch the interface
|
| 45 |
+
iface.launch(show_api=False) # Start the Gradio UI; hide the API schema part of the interface
|