File size: 2,780 Bytes
dc00aa1 c04b26f dc00aa1 c04b26f dc00aa1 c04b26f dc00aa1 c04b26f dc00aa1 c04b26f dc00aa1 c04b26f dc00aa1 8f06561 dc00aa1 8f06561 dc00aa1 8f06561 dc00aa1 |
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 |
import os
import subprocess
import openai
import gradio as gr
import requests
from gtts import gTTS
API_URL = "https://api-inference.huggingface.co/models/lyimo/whisper-small-sw2"
headers = {"Authorization": "Bearer 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"}
openai.api_key = "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"
def query(filename):
with open(filename, "rb") as f:
data = f.read()
response = requests.post(API_URL, headers=headers, data=data)
return response.json()
def transcribe(audio):
output = query(audio)
return output["text"]
def generate_response(transcribed_text):
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=[
{"role": "system", "content": "Wewe ni mtaalamu wa viazi lishe na utajibu maswali yote kwa kiswahili tu!"},
{"role": "user", "content": "Mambo vipi?"},
{"role": "assistant", "content": """Salama je una swali lolote kuhusu viazi lishe?"""},
{"role": "user", "content": "nini maana ya Viazi lishe?"},
{"role": "assistant", "content": """ viazi lishe ni Viazi vitamu vyenye rangi ya karoti kwa ndani ambavyo vina vitamin A kwa wingi"""},
{"role": "user", "content": "nini matumizi ya viazi lishe?"},
{"role": "assistant", "content": """ viazi lishe vinaweza kutengenezea chakula kama Keki,
Maandazi, Kalimati na tambi: Ukisaga unga wa viazi lishe,
unaweza kutumika kupika vyakula ambavyo huwa watu
hutumia unga wa ngano kupika, unga wa viazi lishe una
virutubisho vingi zaidi kuliko unga wa ngano na
ukitumika kupikia vyakula tajwa hapo juu watumiaji
watakuwa wanakula vyakula vyenye virutubisho Zaidi."""},
{"role": "user", "content": transcribed_text},
]
)
return response['choices'][0]['message']['content']
def inference(text):
output_file = "tts_output.wav"
tts = gTTS(text, lang="sw")
tts.save(output_file)
return output_file
def process_audio_and_respond(audio):
text = transcribe(audio)
response_text = generate_response(text)
output_file = inference(response_text)
return response_text, output_file
demo = gr.Interface(
process_audio_and_respond,
gr.inputs.Audio(source="microphone", type="filepath", label="Bonyeza kitufe cha kurekodi na uliza swali lako"),
[gr.outputs.Textbox(label="Jibu (kwa njia ya maandishi)"), gr.outputs.Audio(type="filepath", label="Jibu kwa njia ya sauti (Bofya kusikiliza Jibu)")],
title="Mtaalamu wa Viazi Lishe",
description="Uliza Mtaalamu wetu swali lolote Kuhusu viazi Lishe",
theme="compact",
layout="vertical",
allow_flagging=False,
live=True,
)
demo.launch() |