Spaces:
Build error
Build error
File size: 2,860 Bytes
5b5d4af fe33c17 11f7102 5b5d4af cfc38a8 fe33c17 5b5d4af fe33c17 006d225 fe33c17 6c3ad82 71471a7 6c3ad82 fe33c17 7f1ab16 fe33c17 ea180c8 71471a7 ea180c8 71471a7 ea180c8 71471a7 7f1ab16 a1f8c85 71471a7 fe33c17 ea180c8 6c3ad82 03606ff 6c3ad82 03606ff 6c3ad82 fe33c17 6c3ad82 fe33c17 ea180c8 fe33c17 d325a90 5b5d4af 71471a7 5b5d4af 9ca6873 5b5d4af fe33c17 5b5d4af |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 |
import os
#import numpy as np
import gradio as gr
import whisper
import requests
import tempfile
from neon_tts_plugin_coqui import CoquiTTS
# Whisper: Speech-to-text
model = whisper.load_model("base")
# The LLM : Bloom
API_URL = "https://api-inference.huggingface.co/models/bigscience/bloom"
HF_TOKEN = os.environ["HF_TOKEN"]
headers = {"Authorization": f"Bearer {HF_TOKEN}"}
#en, fr, esp, arb, hn, portu, Indonesian, Vietnamese, Chinese, tamil, telugu, bengali
# Text-to-Speech
LANGUAGES = list(CoquiTTS.langs.keys())
print(f"Languages for Coqui are: {LANGUAGES}")
coquiTTS = CoquiTTS()
# Whisper - speeech-to-text
def whisper_stt(audio):
# 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)
# print the recognized text
print(f"transcript is : {result.text}")
return result.text
# Driver function
def fun_engine(audio) :
text1 = whisper_stt(audio)
#text1 = model.transcribe(audio)["text"]
text2 = lang_model_response(text1)
speech = tts(text, 'en')
return text1, text2, speech
# LLM - Bloom Response
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": 0.90, #0.90 default
"max_new_tokens": 64,
"temperature": 1.1, #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.split(".")[1]
print(f"Final response after splits is: {solution}")
return solution
# Coqui - Text-to-Speech
def tts(text, language):
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
coquiTTS.get_tts(text, fp, speaker = {"language" : language})
return fp.name
gr.Interface(
title = 'Testing Whisper',
fn=fun_engine,
inputs=[
gr.Audio(source="microphone", type="filepath"), #streaming = True,
# "state"
],
outputs=[
"textbox", "textbox", "audio",
],
live=True).launch()
|