Update app.py
Browse files
app.py
CHANGED
@@ -1,85 +1,111 @@
|
|
1 |
-
import streamlit as st
|
2 |
-
from transformers import WhisperProcessor, WhisperForConditionalGeneration, RagTokenizer, RagRetriever, RagSequenceForGeneration
|
3 |
-
import torch
|
4 |
-
import soundfile as sf
|
5 |
-
import librosa
|
6 |
-
from moviepy.editor import VideoFileClip
|
7 |
-
import os
|
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 |
-
transcription
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from transformers import WhisperProcessor, WhisperForConditionalGeneration, RagTokenizer, RagRetriever, RagSequenceForGeneration
|
3 |
+
import torch
|
4 |
+
import soundfile as sf
|
5 |
+
import librosa
|
6 |
+
from moviepy.editor import VideoFileClip
|
7 |
+
import os
|
8 |
+
import tempfile
|
9 |
+
import logging
|
10 |
+
|
11 |
+
# Set up logging
|
12 |
+
logging.basicConfig(level=logging.INFO)
|
13 |
+
logger = logging.getLogger(__name__)
|
14 |
+
|
15 |
+
# Load Whisper base model and processor
|
16 |
+
whisper_model_name = "openai/whisper-base"
|
17 |
+
whisper_processor = WhisperProcessor.from_pretrained(whisper_model_name)
|
18 |
+
whisper_model = WhisperForConditionalGeneration.from_pretrained(whisper_model_name)
|
19 |
+
|
20 |
+
# Load RAG sequence model and tokenizer
|
21 |
+
rag_model_name = "facebook/rag-sequence-nq"
|
22 |
+
rag_tokenizer = RagTokenizer.from_pretrained(rag_model_name)
|
23 |
+
|
24 |
+
# Try to load RagRetriever with trust_remote_code=True
|
25 |
+
try:
|
26 |
+
rag_retriever = RagRetriever.from_pretrained(
|
27 |
+
rag_model_name,
|
28 |
+
index_name="exact",
|
29 |
+
use_dummy_dataset=True,
|
30 |
+
trust_remote_code=True
|
31 |
+
)
|
32 |
+
logger.info("Successfully loaded RagRetriever with trust_remote_code=True")
|
33 |
+
except ValueError as e:
|
34 |
+
logger.error(f"Error loading RagRetriever: {e}")
|
35 |
+
st.error(f"Error loading RagRetriever: {e}")
|
36 |
+
|
37 |
+
rag_model = RagSequenceForGeneration.from_pretrained(rag_model_name, retriever=rag_retriever)
|
38 |
+
|
39 |
+
def transcribe_audio(audio_path, language="ru"):
|
40 |
+
speech, rate = librosa.load(audio_path, sr=16000)
|
41 |
+
inputs = whisper_processor(speech, return_tensors="pt", sampling_rate=16000)
|
42 |
+
input_features = whisper_processor.feature_extractor(speech, return_tensors="pt", sampling_rate=16000).input_features
|
43 |
+
predicted_ids = whisper_model.generate(input_features, forced_decoder_ids=whisper_processor.get_decoder_prompt_ids(language=language, task="translate"))
|
44 |
+
transcription = whisper_processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
|
45 |
+
return transcription
|
46 |
+
|
47 |
+
def translate_and_summarize(text):
|
48 |
+
inputs = rag_tokenizer(text, return_tensors="pt")
|
49 |
+
input_ids = inputs["input_ids"]
|
50 |
+
attention_mask = inputs["attention_mask"]
|
51 |
+
outputs = rag_model.generate(input_ids=input_ids, attention_mask=attention_mask)
|
52 |
+
return rag_tokenizer.batch_decode(outputs, skip_special_tokens=True)
|
53 |
+
|
54 |
+
def extract_audio_from_video(video_path, output_audio_path):
|
55 |
+
video_clip = VideoFileClip(video_path)
|
56 |
+
audio_clip = video_clip.audio
|
57 |
+
if audio_clip is not None:
|
58 |
+
audio_clip.write_audiofile(output_audio_path)
|
59 |
+
return output_audio_path
|
60 |
+
else:
|
61 |
+
return None
|
62 |
+
|
63 |
+
st.title("Audio and Video Transcription & Summarization")
|
64 |
+
|
65 |
+
# Audio Upload Section
|
66 |
+
st.header("Upload an Audio File")
|
67 |
+
audio_file = st.file_uploader("Choose an audio file...", type=["wav", "mp3", "m4a"])
|
68 |
+
|
69 |
+
if audio_file is not None:
|
70 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
|
71 |
+
tmp_file.write(audio_file.getbuffer())
|
72 |
+
audio_path = tmp_file.name
|
73 |
+
|
74 |
+
st.audio(audio_file)
|
75 |
+
st.write("Transcribing audio...")
|
76 |
+
try:
|
77 |
+
transcription = transcribe_audio(audio_path)
|
78 |
+
st.write("Transcription:", transcription)
|
79 |
+
|
80 |
+
st.write("Translating and summarizing...")
|
81 |
+
summary = translate_and_summarize(transcription)
|
82 |
+
st.write("Translated Summary:", summary)
|
83 |
+
except Exception as e:
|
84 |
+
st.error(f"An error occurred: {e}")
|
85 |
+
|
86 |
+
# Video Upload Section
|
87 |
+
st.header("Upload a Video File")
|
88 |
+
video_file = st.file_uploader("Choose a video file...", type=["mp4", "mkv", "avi", "mov"])
|
89 |
+
|
90 |
+
if video_file is not None:
|
91 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as tmp_file:
|
92 |
+
tmp_file.write(video_file.getbuffer())
|
93 |
+
video_path = tmp_file.name
|
94 |
+
|
95 |
+
st.video(video_file)
|
96 |
+
st.write("Extracting audio from video...")
|
97 |
+
audio_path = extract_audio_from_video(video_path, tempfile.NamedTemporaryFile(delete=False, suffix=".wav").name)
|
98 |
+
|
99 |
+
if audio_path is not None:
|
100 |
+
st.write("Transcribing audio...")
|
101 |
+
try:
|
102 |
+
transcription = transcribe_audio(audio_path)
|
103 |
+
st.write("Transcription:", transcription)
|
104 |
+
|
105 |
+
st.write("Translating and summarizing...")
|
106 |
+
summary = translate_and_summarize(transcription)
|
107 |
+
st.write("Translated Summary:", summary)
|
108 |
+
except Exception as e:
|
109 |
+
st.error(f"An error occurred: {e}")
|
110 |
+
else:
|
111 |
+
st.write("No audio track found in the video file.")
|