Spaces:
Runtime error
Runtime error
import whisper | |
import gradio as gr | |
import openai | |
import os | |
openai.api_key = os.environ["OPENAI_API_KEY"] | |
model = whisper.load_model("small") | |
def transcribe(audio): | |
model = whisper.load_model("base") | |
result = model.transcribe(audio) | |
return result["text"] | |
# def transcribe(audio): | |
# #time.sleep(3) | |
# # load audio and pad/trim it to fit 30 seconds | |
# audio = whisper.load_audio(audio) | |
# audio = whisper.pad_or_trim(audio) | |
# # make log-Mel spectrogram and move to the same device as the model | |
# mel = whisper.log_mel_spectrogram(audio).to(model.device) | |
# # detect the spoken language | |
# _, probs = model.detect_language(mel) | |
# print(f"Detected language: {max(probs, key=probs.get)}") | |
# # decode the audio | |
# options = whisper.DecodingOptions(fp16 = False) | |
# result = whisper.decode(model, mel, options) | |
# return result.text | |
def process_text(input_text): | |
# Apply your function here to process the input text | |
output_text = input_text.upper() | |
return output_text | |
def get_completion(prompt, model='gpt-3.5-turbo'): | |
messages = [ | |
{"role": "system", "content": """You are a world class nurse practitioner. You are provided with the transcription of a patient's recording in a non-English language prior to doctor's visit. \ | |
Extract the following information from the transcription, replace curly brackets with relevant extracted information, and present in English as follows, one category per line: \ | |
Demographic information: {name, age, gender, address, phone number} | |
Medical history: {chronic health conditions, any past surgery, any hospitalization, current medications} | |
Symptoms: {current symptoms, when did they start, how did they progress} | |
Allergies: {any known allergies, any allergic reaction to medications} | |
Family history: {any family members with chronic health condition, anyone in the family with a hereditary condition?} | |
Lifestyle factors: {typical diet, how often you exercise, smoking, drinking alcohol} | |
Psychosocial factors: {stress, anxiety, any mental health condition} | |
Review of systems: {any issues with vision or hearing, digestive issues, any problems with skin or nails, any problems with joints or muscles} | |
All information in the report needs to be in English only. | |
Only use the information from the provided transcription. Do not make up stuff. If information is not available just put "N/A" next to the relevant line. | |
""" | |
}, | |
{"role": "user", "content": prompt} | |
] | |
response = openai.ChatCompletion.create( | |
model = model, | |
messages = messages, | |
temperature = 0, | |
) | |
return response.choices[0].message['content'] | |
with gr.Blocks() as demo: | |
gr.Markdown(""" | |
# Meet your doctor <br> | |
This is to make life of non-English speaking patients easier. | |
Describe your complaints and symptoms in your native language , have it emailed to your doctor prior to your visit. | |
Information that is useful to include: your name, age, gender, address, phone number, medical history, symptoms, allergies, family medical history, lifestyle factors. | |
Have it all recorded, transcribed, and presented in a standard form. | |
""") | |
title = "Chat with NP" | |
audio = gr.Audio(source="microphone", type="filepath") | |
b1 = gr.Button("Transcribe audio") | |
b2 = gr.Button("Prepare a report in English") | |
b3 = gr.Button("Email report to your doctor") | |
text1 = gr.Textbox(lines=5) | |
text2 = gr.Textbox(lines=5) | |
prompt = text1 | |
b1.click(transcribe, inputs=audio, outputs=text1) | |
b2.click(get_completion, inputs=text1, outputs=text2) | |
# b1.click(transcribe, inputs=audio, outputs=text1) | |
# b2.click(get_completion, inputs=prompt, outputs=text2) | |
demo.launch() | |
#demo.launch(share=True, auth=("username", "password")) | |
# In this example, the process_text function just converts the input text to uppercase, but you can replace it with your desired function. The Gradio Blocks interface will have two buttons: "Transcribe audio" and "Process text". The first button transcribes the audio and fills the first textbox, and the second button processes the text from the first textbox and fills the second textbox. | |
# gr.Interface( | |
# title = 'OpenAI Whisper ASR Gradio Web UI', | |
# fn=transcribe, | |
# inputs=[ | |
# gr.inputs.Audio(source="microphone", type="filepath") | |
# ], | |
# outputs=[ | |
# "textbox" | |
# ], | |
# live=True).launch() | |