Spaces:
Runtime error
Runtime error
Init
Browse files- .gitignore +2 -0
- app.ipynb +92 -0
- app.py +58 -0
- gradio_article.md +16 -0
- requirements.txt +9 -0
.gitignore
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
tmp.ipynb
|
2 |
+
tmp.mp3
|
app.ipynb
ADDED
@@ -0,0 +1,92 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": 1,
|
6 |
+
"metadata": {},
|
7 |
+
"outputs": [
|
8 |
+
{
|
9 |
+
"name": "stderr",
|
10 |
+
"output_type": "stream",
|
11 |
+
"text": [
|
12 |
+
"playsound is relying on another python subprocess. Please use `pip install pygobject` if you want playsound to run more efficiently.\n"
|
13 |
+
]
|
14 |
+
}
|
15 |
+
],
|
16 |
+
"source": [
|
17 |
+
"import openai\n",
|
18 |
+
"from playsound import playsound\n",
|
19 |
+
"from gtts import gTTS\n",
|
20 |
+
"import speech_recognition as sr\n",
|
21 |
+
"import gradio as gr\n",
|
22 |
+
"\n",
|
23 |
+
"openai.api_key = \"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\"\n",
|
24 |
+
"# will hide the api key:\n",
|
25 |
+
"# import openai_secret_manager\n",
|
26 |
+
"# assert \"openai\" in openai_secret_manager.get_services()\n",
|
27 |
+
"# secrets = openai_secret_manager.get_secret(\"openai\")\n",
|
28 |
+
"# openai.api_key = secrets[\"api_key\"]"
|
29 |
+
]
|
30 |
+
},
|
31 |
+
{
|
32 |
+
"cell_type": "code",
|
33 |
+
"execution_count": 2,
|
34 |
+
"metadata": {},
|
35 |
+
"outputs": [],
|
36 |
+
"source": [
|
37 |
+
"def generate_response(prompt):\n",
|
38 |
+
" prompt = (f\"{prompt}\")\n",
|
39 |
+
"\n",
|
40 |
+
" response = openai.ChatCompletion.create(\n",
|
41 |
+
" model=\"gpt-3.5-turbo\",\n",
|
42 |
+
" messages=[\n",
|
43 |
+
" {\"role\": \"user\", \"content\": f\"{prompt}\"},\n",
|
44 |
+
" ])\n",
|
45 |
+
"\n",
|
46 |
+
" message = response.choices[0]['message']['content']\n",
|
47 |
+
" return message"
|
48 |
+
]
|
49 |
+
},
|
50 |
+
{
|
51 |
+
"cell_type": "code",
|
52 |
+
"execution_count": 8,
|
53 |
+
"metadata": {},
|
54 |
+
"outputs": [
|
55 |
+
{
|
56 |
+
"data": {
|
57 |
+
"text/plain": [
|
58 |
+
"\"I apologize, but as a language model AI, I don't have access to real-time information. Could you please check the time on your device or ask a nearby clock?\""
|
59 |
+
]
|
60 |
+
},
|
61 |
+
"execution_count": 8,
|
62 |
+
"metadata": {},
|
63 |
+
"output_type": "execute_result"
|
64 |
+
}
|
65 |
+
],
|
66 |
+
"source": [
|
67 |
+
"generate_response(\"What time is it?\")"
|
68 |
+
]
|
69 |
+
}
|
70 |
+
],
|
71 |
+
"metadata": {
|
72 |
+
"kernelspec": {
|
73 |
+
"display_name": "base",
|
74 |
+
"language": "python",
|
75 |
+
"name": "python3"
|
76 |
+
},
|
77 |
+
"language_info": {
|
78 |
+
"codemirror_mode": {
|
79 |
+
"name": "ipython",
|
80 |
+
"version": 3
|
81 |
+
},
|
82 |
+
"file_extension": ".py",
|
83 |
+
"mimetype": "text/x-python",
|
84 |
+
"name": "python",
|
85 |
+
"nbconvert_exporter": "python",
|
86 |
+
"pygments_lexer": "ipython3",
|
87 |
+
"version": "3.8.13"
|
88 |
+
}
|
89 |
+
},
|
90 |
+
"nbformat": 4,
|
91 |
+
"nbformat_minor": 2
|
92 |
+
}
|
app.py
ADDED
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import openai
|
2 |
+
from playsound import playsound
|
3 |
+
from gtts import gTTS
|
4 |
+
import speech_recognition as sr
|
5 |
+
import gradio as gr
|
6 |
+
|
7 |
+
openai.api_key = api_key
|
8 |
+
|
9 |
+
def generate_response(prompt):
|
10 |
+
prompt = (f"{prompt}")
|
11 |
+
|
12 |
+
response = openai.ChatCompletion.create(
|
13 |
+
model="gpt-3.5-turbo",
|
14 |
+
messages=[
|
15 |
+
{"role": "user", "content": f"{prompt}"},
|
16 |
+
])
|
17 |
+
|
18 |
+
message = response.choices[0]['message']['content']
|
19 |
+
return message
|
20 |
+
|
21 |
+
r = sr.Recognizer()
|
22 |
+
from pydub import AudioSegment
|
23 |
+
def transcribe(audio, lang):
|
24 |
+
with sr.AudioFile(audio) as source: audio = r.record(source)
|
25 |
+
text = r.recognize_google(audio, language=lang)
|
26 |
+
text = generate_response(text)
|
27 |
+
tts = gTTS(text=text, lang=lang)
|
28 |
+
out = "tmp.mp3"
|
29 |
+
tts.save(out)
|
30 |
+
return out
|
31 |
+
|
32 |
+
with open('gradio_article.md') as f:
|
33 |
+
article = f.read()
|
34 |
+
|
35 |
+
interface_options = {
|
36 |
+
"title": "Smart GPT",
|
37 |
+
"description": "Let's have a chat! Talk to me, and I'll respond in a jiffy",
|
38 |
+
"article": article,
|
39 |
+
"layout": "horizontal",
|
40 |
+
"theme": "default",
|
41 |
+
}
|
42 |
+
|
43 |
+
inputs = gr.Audio(source="microphone", type="filepath")
|
44 |
+
outputs = "audio"
|
45 |
+
|
46 |
+
lang = gr.Dropdown(choices=["en", "vi", "nl"], value="en")
|
47 |
+
|
48 |
+
gr.Interface(fn=transcribe, inputs=[inputs, lang], outputs=outputs, live=True,
|
49 |
+
**interface_options).launch()
|
50 |
+
|
51 |
+
# TODO
|
52 |
+
# Custom voice
|
53 |
+
# VALL-E
|
54 |
+
# https://cloud.google.com/text-to-speech/custom-voice/docs/quickstart
|
55 |
+
# Mozilla TTS
|
56 |
+
# OpenSeq2Seq
|
57 |
+
# Best VN: Vbee, FPT
|
58 |
+
# Elevenlabs for English
|
gradio_article.md
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
## Description
|
2 |
+
Hey there,
|
3 |
+
|
4 |
+
I made this cool app, mainly for my little son. It's a fun voice-controlled thing where he can chat with it and get spoken answers. For example, he can ask it to tell a story or find out stuff he's curious about.
|
5 |
+
|
6 |
+
Here's how it works:
|
7 |
+
|
8 |
+
I used a bunch of libraries like openai, playsound, gtts, speech_recognition, and gradio. The app records what you say into a microphone and sends it to OpenAI for answers. Then, it turns those answers into speech and plays them back to you.
|
9 |
+
|
10 |
+
There's a "transcribe" function that takes audio and a language code, listens to what you say, asks OpenAI for answers, turns those answers into speech, and stores it temporarily. Then it gives you the path to the spoken response.
|
11 |
+
|
12 |
+
The app also has a simple user interface made with gradio. You can pick a language and talk into the microphone. The app transcribes your words, gets answers, and plays them back to you.
|
13 |
+
|
14 |
+
I'm thinking of adding even more fun stuff like custom voices from VALL-E, Mozilla TTS, OpenSeq2Seq, Vbee, FPT, or Elevenlabs. So I can train my own voice to use for the generated response. Isn't it interesting?
|
15 |
+
|
16 |
+
This is a fun and easy way to have voice conversations and learn new things. So, it's not just for my son; anyone can enjoy it!
|
requirements.txt
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
gradio==3.39.0
|
2 |
+
gTTS==2.3.1
|
3 |
+
openai==0.27.4
|
4 |
+
playsound==1.3.0
|
5 |
+
SpeechRecognition==3.9.0
|
6 |
+
transformers
|
7 |
+
torch
|
8 |
+
# pygobject
|
9 |
+
ffmpeg
|