Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,20 +1,52 @@
|
|
1 |
import gradio as gr
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
speech_to_text = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
|
6 |
-
#text_generation = pipeline("text-generation", model="microsoft/Phi-3-mini-128k-instruct", trust_remote_code = True)
|
7 |
-
#text_to_speech = pipeline("text-to-speech", model="kakao-enterprise/vits-ljs")
|
8 |
|
9 |
def alexa(audio):
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
return text
|
15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
gr.Interface(
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
outputs = gr.Textbox(), live=True).launch()
|
|
|
1 |
import gradio as gr
|
2 |
+
import torch
|
3 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
|
|
|
|
|
|
|
|
|
4 |
|
5 |
def alexa(audio):
|
6 |
+
converted_text = speech_to_text(audio)
|
7 |
+
generated_text = text_generation(text)
|
8 |
+
speech = text_to_speech(generated_text)
|
9 |
+
return speech
|
10 |
+
|
11 |
+
def speech_to_text(audio):
|
12 |
+
audio_to_text = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
|
13 |
+
if audio == None:
|
14 |
+
raise gr.Error("Please, submit an audio file")
|
15 |
+
else:
|
16 |
+
text = audio_to_text(audio)["text"]
|
17 |
return text
|
18 |
|
19 |
+
def text_generation(text):
|
20 |
+
torch.random.manual_seed(0)
|
21 |
+
|
22 |
+
model = AutoModelForCausalLM.from_pretrained("microsoft/Phi-3-mini-128k-instruct",
|
23 |
+
device_map="cuda",
|
24 |
+
torch_dtype="auto",
|
25 |
+
trust_remote_code=True, )
|
26 |
+
tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3-mini-128k-instruct")
|
27 |
+
|
28 |
+
messages = [
|
29 |
+
{"role": "user", "content": text
|
30 |
+
]
|
31 |
+
|
32 |
+
generation_args = {
|
33 |
+
"max_new_tokens": 500,
|
34 |
+
"return_full_text": False,
|
35 |
+
"temperature": 0.0,
|
36 |
+
"do_sample": False,
|
37 |
+
}
|
38 |
+
|
39 |
+
text_gen= pipeline("text-generation", model="microsoft/Phi-3-mini-128k-instruct", trust_remote_code = True)
|
40 |
+
response = text_gen(messages, **generation_args)
|
41 |
+
return response[0]["generated_text"]
|
42 |
+
|
43 |
+
def text_to_speech(text):
|
44 |
+
text_to_audio = pipeline("text-to-speech", model="kakao-enterprise/vits-ljs")
|
45 |
+
narrated_text = text_to_audio(text)
|
46 |
+
return (narrated_text["sampling_rate"], narrated_text["audio"][0] )
|
47 |
+
|
48 |
+
|
49 |
gr.Interface(
|
50 |
+
fn=alexa,
|
51 |
+
inputs=gr.Audio(type="filepath"),
|
52 |
+
outputs=[gr.Audio(label="Narration", type="numpy", autoplay=True)], live=True).launch()
|
|