Update app.py
Browse files
app.py
CHANGED
@@ -1,20 +1,29 @@
|
|
1 |
from transformers import pipeline
|
2 |
import gradio as gr
|
3 |
from gtts import gTTS
|
|
|
4 |
|
5 |
# Load the Whisper model for speech-to-text
|
6 |
pipe = pipeline(model="openai/whisper-small")
|
7 |
|
8 |
# Load the text generation model
|
9 |
-
text_pipe = pipeline("text2text-generation", model="google/flan-t5-base")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
|
11 |
def transcribe(audio):
|
12 |
# Transcribe the audio to text
|
13 |
text = pipe(audio)["text"]
|
14 |
|
15 |
# Generate a response from the transcribed text
|
16 |
-
lm_response = text_pipe(text)[0]["generated_text"]
|
17 |
-
|
18 |
# Convert the response text to speech
|
19 |
tts = gTTS(lm_response, lang='ko')
|
20 |
|
|
|
1 |
from transformers import pipeline
|
2 |
import gradio as gr
|
3 |
from gtts import gTTS
|
4 |
+
import openai
|
5 |
|
6 |
# Load the Whisper model for speech-to-text
|
7 |
pipe = pipeline(model="openai/whisper-small")
|
8 |
|
9 |
# Load the text generation model
|
10 |
+
# text_pipe = pipeline("text2text-generation", model="google/flan-t5-base")
|
11 |
+
|
12 |
+
def generate_gpt_response(text):
|
13 |
+
response = openai.Completion.create(
|
14 |
+
engine="text-davinci-003", # Use the appropriate GPT-3 engine
|
15 |
+
prompt=text,
|
16 |
+
max_tokens=150
|
17 |
+
)
|
18 |
+
return response.choices[0].text.strip()
|
19 |
|
20 |
def transcribe(audio):
|
21 |
# Transcribe the audio to text
|
22 |
text = pipe(audio)["text"]
|
23 |
|
24 |
# Generate a response from the transcribed text
|
25 |
+
# lm_response = text_pipe(text)[0]["generated_text"]
|
26 |
+
lm_response = generate_gpt_response(text)
|
27 |
# Convert the response text to speech
|
28 |
tts = gTTS(lm_response, lang='ko')
|
29 |
|