ysharma's picture
ysharma HF staff
q
6c3ad82
raw history blame
No virus
1.74 kB
import os
import numpy as np
import gradio as gr
import whisper
model = whisper.load_model("base")
##Bloom
API_URL = "https://api-inference.huggingface.co/models/bigscience/bloom"
HF_TOKEN = os.environ["HF_TOKEN"]
headers = {"Authorization": f"Bearer {HF_TOKEN}"}
def fun(audio) : #, state=''):
text1 = model.transcribe(audio)["text"]
text2 = lang_model_response(text)
return text1, text2
def lang_model_response(prompt):
print(f"*****Inside meme_generate - Prompt is :{prompt}")
if len(prompt) == 0:
prompt = """Can you help me please?"""
json_ = {"inputs": prompt,
"parameters":
{
"top_p": top_p, #0.90 default
"max_new_tokens": 64,
"temperature": temp, #1.1 default
"return_full_text": True,
"do_sample": True,
},
"options":
{"use_cache": True,
"wait_for_model": True,
},}
response = requests.post(API_URL, headers=headers, json=json_)
print(f"Response is : {response}")
output = response.json()
print(f"output is : {output}")
output_tmp = output[0]['generated_text']
print(f"output_tmp is: {output_tmp}")
solution = output_tmp[0] #output_tmp.split("\nQ:")[0]
print(f"Final response after splits is: {solution}")
#meme_image, new_prompt = write_on_image(solution)
return solution
def fun1(audio, state=''):
text = model.transcribe(audio)["text"]
state += text + " "
return state, state
gr.Interface(
title = 'Testing Whisper',
fn=fun,
inputs=[
gr.Audio(source="microphone", type="filepath"), #streaming = True,
# "state"
],
outputs=[
"textbox", "textbox"
],
live=True).launch()