Spaces:
Running
Running
Replace the code
Browse files- .gitattributes +6 -5
- .gitignore +2 -1
- README.md +8 -178
- app.py +126 -87
- inference.py +0 -68
- inference_gpu.py +0 -69
- inference_timestamps.py +0 -86
- mer_lviv_interview.wav +0 -3
- requirements-dev.txt +1 -0
- requirements.txt +8 -9
- tsn_2.wav β sample_1.wav +2 -2
- short_1.wav β sample_2.wav +2 -2
- tsn.wav β sample_3.wav +2 -2
- long_1.wav β sample_4.wav +2 -2
- sample_5.wav +3 -0
- sample_6.wav +3 -0
.gitattributes
CHANGED
@@ -1,5 +1,6 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
|
|
|
1 |
+
sample_1.wav filter=lfs diff=lfs merge=lfs -text
|
2 |
+
sample_2.wav filter=lfs diff=lfs merge=lfs -text
|
3 |
+
sample_3.wav filter=lfs diff=lfs merge=lfs -text
|
4 |
+
sample_4.wav filter=lfs diff=lfs merge=lfs -text
|
5 |
+
sample_5.wav filter=lfs diff=lfs merge=lfs -text
|
6 |
+
sample_6.wav filter=lfs diff=lfs merge=lfs -text
|
.gitignore
CHANGED
@@ -1,2 +1,3 @@
|
|
1 |
.idea/
|
2 |
-
.venv
|
|
|
|
1 |
.idea/
|
2 |
+
.venv/
|
3 |
+
flagged/
|
README.md
CHANGED
@@ -1,184 +1,14 @@
|
|
1 |
-
|
2 |
-
title: Wav2vec2 Ukrainian with Timestamps
|
3 |
-
emoji: πΊπ¦
|
4 |
-
colorFrom: blue
|
5 |
-
colorTo: yellow
|
6 |
-
sdk: gradio
|
7 |
-
app_file: app.py
|
8 |
-
pinned: false
|
9 |
-
---
|
10 |
|
|
|
11 |
|
12 |
-
|
|
|
13 |
|
14 |
-
|
15 |
-
- The model with better News LM: https://huggingface.co/Yehor/wav2vec2-xls-r-1b-uk-with-news-lm
|
16 |
|
17 |
-
|
18 |
|
19 |
-
|
20 |
-
|
21 |
-
Create a virtualenv:
|
22 |
-
|
23 |
-
```bash
|
24 |
-
pipenv install
|
25 |
-
pipenv shell
|
26 |
-
```
|
27 |
-
|
28 |
-
Install deps:
|
29 |
-
|
30 |
-
```bash
|
31 |
-
pip install https://github.com/huggingface/transformers/archive/refs/tags/v4.16.2.zip
|
32 |
-
pip install https://github.com/kpu/kenlm/archive/master.zip
|
33 |
-
|
34 |
-
pip install torch==1.9.1 torchaudio==0.9.1 pyctcdecode==0.3.0
|
35 |
-
```
|
36 |
-
|
37 |
-
Run inference:
|
38 |
-
|
39 |
-
```bash
|
40 |
-
python inference.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-lm --path_files short_1.wav
|
41 |
-
|
42 |
-
# with chunking
|
43 |
-
python inference.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-lm --path_files short_1.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
44 |
-
python inference.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-lm --path_files long_1.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
45 |
-
|
46 |
-
# with chunking on GPU
|
47 |
-
python inference_gpu.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-lm --path_files short_1.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
48 |
-
python inference_gpu.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-lm --path_files long_1.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
49 |
-
|
50 |
-
python inference.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-news-lm --path_files mer_lviv_interview.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
51 |
-
python inference_gpu.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-news-lm --path_files mer_lviv_interview.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
52 |
-
|
53 |
-
python inference.py --model_id Yehor/wav2vec2-xls-r-1b-uk-with-lm --path_files tsn.wav,tsn_2.wav --chunk_length_s 10 --stride_length_s_l 4 --stride_length_s_r 2
|
54 |
-
```
|
55 |
-
|
56 |
-
NOTE: Do the inference process for long files with chunking.
|
57 |
-
|
58 |
-
---
|
59 |
-
|
60 |
-
short_1.wav:
|
61 |
-
|
62 |
-
```
|
63 |
-
ΠΏΠ°Π½Π° ΡΠΏΠΎΠ»ΡΡΠ΅Π½Ρ ΡΡΠ°ΡΠΈ Π½Π°Π΄ Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½ΠΈΠΉ ΠΏΠ°ΡΡΠ½Π΅Ρ ΠΎΠ΄Π½Π°ΠΊ Ρ ΡΡΠ·Π½ΠΈΡΡΡΡΠ°ΡΠΈ ΠΌΠ°ΡΡΡ ΡΠΏΠ΅ΡΡΠ°Π»ΡΠ½ΠΈΠΉ Π·Π°ΠΊΠΎΠ½ ΡΠΊΠΈΠΉ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ°Ρ ΡΠΊΡΠΎ ΠΊΠΈΡΠ°ΠΉ Π½Π°ΠΏΠ°Π΄ΠΈ Π½Π° ΡΠ°ΠΉΠ²Π°Π½Ρ Π°ΠΌΠ΅ΡΠΈΠΊΠ°Π½ΡΡΠΊΠΈΠΉ Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ ΠΌΠ°ΡΡΡ ΠΉΠΎΠ³ΠΎ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Ρ Π³ΡΠΈ
|
64 |
-
```
|
65 |
-
|
66 |
-
short_1.wav (with better News LM):
|
67 |
-
|
68 |
-
```
|
69 |
-
Π°Π½Ρ ΡΠΏΠΎΠ»ΡΡΠ΅Π½Ρ ΡΡΠ°ΡΠΈ Π½Π°Π΄ Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½ΠΈΠΉ ΠΏΠ°ΡΡΠ½Π΅Ρ ΠΎΠ΄Π½Π°ΠΊ Ρ ΡΡΠ·Π½ΠΈΡΡ ΡΡΠ°ΡΠΈ ΠΌΠ°ΡΡΡ ΡΠΏΠ΅ΡΡΠ°Π»ΡΠ½ΠΈΠΉ Π·Π°ΠΊΠΎΠ½ ΡΠΊΠΈΠΉ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ°Ρ ΡΠΊΡΠΎ ΠΊΠΈΡΠ°ΠΉ Π½Π°ΠΏΠ°Π΄Π΅ Π½Π° ΡΠ°ΠΉΠ²Π°Π½Ρ Π°ΠΌΠ΅ΡΠΈΠΊΠ°Π½ΡΡΠΊΠΈΠΉ Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ ΠΌΠ°ΡΡΡ ΠΉΠΎΠ³ΠΎ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ ΡΠ³Π΅ΡΠΈ
|
70 |
-
```
|
71 |
-
|
72 |
-
long_1.wav:
|
73 |
-
|
74 |
-
```
|
75 |
-
ΡΠ΅ΡΡΠ΅ ΡΠΈ Π΄ΠΈΠ²ΠΎΠ²ΠΈΠΆΠ½ΠΈ ΠΏΠΎΡΡΡΡΠ½ΠΎΠΊ ΠΌΡΠ»ΡΠΉΠΎΠ½ΠΈ Π»ΡΠ΄Π΅ΠΉ ΡΠ°ΠΊΡΠΈΡΠ½ΠΎ Π² ΠΏΡΡΠΌΠΎΠΌΡ Π΅ΡΡΡΡ Π²ΠΆΠ΅ ΡΡΠΈ Π΄ΠΎΠ±ΠΈ ΡΠΏΠΎΡΡΠ΅ΡΡΠ³Π°ΡΡΡ Π·Π° ΡΠΏΡΠΎΠ±Π°ΠΌΠΈΠ°ΠΌΠ΅ΡΠΎΠΊΠ°Π½ΡΡΠΊΠΈΡ
ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΠΈΠ² Π΄ΡΡΡΡΠΈΡΠΊΠΎΠ»ΠΎΠ΄Ρ Π·Π° ΠΏΡΡΠΈΡΡΡΠ½Π΅ Ρ
Π»ΠΎΠΏΡ Π΄ΠΎΡΡ Π½Π΅ Π·ΡΠΎΠ·ΡΠΌΡΠ»ΠΎ ΡΠΈ Π²Π΄Π°ΡΡΡΠ΄ΡΡΡΠ°ΡΠΈ ΠΉΠΎΠ³ΠΎ Π· ΡΡΠΈΠ΄ΡΡΡΠΈ ΠΌΠ΅ΡΡΠΎΠ²ΠΎΠ³ΠΎ ΠΏΡΠΎΠ²Π°Π» ΠΆΠΈΠ²ΠΈΠΌ ΠΏΡΠΎ Π½Π°Π΄Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎ ΡΠΊΠ»Π°Π΄Π½Ρ ΠΎΠΏΠ΅ΡΠ°ΡΡΡ ΡΠΎ ΡΡΠΈΠ²Π°Ρ Π² ΡΡ ΠΌΠΈΡΡ Ρ Π½Π° Π΅ΡΠ°ΡΡΡΠΊΠΊΡΠ»ΠΎΡΠ· ΡΠΊΠΈΠΉ ΠΏΡΠΎΠ²Π°Π»ΠΈΠ²ΡΡ ΠΏΡΡΠΈΡΡΡΠ½ΠΈΠΉΡΠ°ΡΠ½ Π»Π΅Π΄Ρ ΠΏΠΎΠΌΡΡΠ½Π° Π΄ΡΡΠ° Π² Π·Π΅ΠΌΠ»Ρ ΠΌΠ΅Π½ΡΠ΅ ΡΡΠΈΠ΄ΡΡΡΠΈ ΡΠ°Π½ΡΠΈΠΌΠ΅ΡΡΡΠ²Ρ Π΄ΡΠ°ΠΌΠ΅ΡΡΡ Π°Π»Π΅ Π² Π³Π»ΠΈΠ± Π²ΠΎΠ½Π° ΡΡΠ³Π½Π΅ΡΡΡΡ Π½Π° ΡΡΠΈΠ΄ΡΡΡΡ Π΄Π²Π° ΠΌΠ΅ΡΡΠΎ Π±Π°ΡΡΠΊΠΈ ΡΡΠΊΠ°Π»ΠΈ ΡΠΈΠ½Π° ΠΊΡΠ»ΡΠΊΠ° ΠΎΠ΄ΠΈΠ½ ΠΏΠ΅ΡΠ΅Π΄ ΡΠΈΠΌ ΡΠΊ Π·ΡΠΎΠ·ΡΠΌΡΠ»Π΅ Π²ΡΠ½ ΠΏΡΠ΄ Π·Π΅ΠΌΠ»Π΅Ρ ΠΊΠΎΠ»ΠΈ Π²ΡΠ½ Π·Π½ΠΈΠΊ Ρ ΠΌΠΎΠ»ΠΈΠ»ΠΈΡΡ Π±ΠΎΠ³ΡΠΏΡΠΎΡΠΈΠ»Π° Π°Π±ΠΈ Π°Π»Π°Π³Π·Π±ΠΈΡΡΠ³ ΠΌΠΎΡΠΈΠ½Π° Ρ ΠΉΠΎΠ³ΠΎ Π΄ΡΡΡΠ°Π»ΠΈ Π· ΠΊΠΎΠ»ΠΎΠ΄ΡΠ·Ρ ΠΆΠΈΠ²ΠΈΠΌ Π³ΠΎΡΠΏΠΎΠ΄ΠΈΡ
Π°ΠΉ ΠΉΠΎΠΌΡ ΡΠ° ΠΌΠ΅Π½ΡΠ΅ Π±ΠΎΠ»ΠΈΡΡ Π² ΡΡΠΉ Π΄ΡΠ»Ρ Ρ ΡΠ°ΠΊ ΡΠΏΠΎΠ΄ΡΠ²Π°ΡΡΠΈΡ ΡΠΎ Ρ ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΠΈ Π²ΡΠ΅ Π²ΠΈΠΉΠ΄Π΅ ΠΉΠΎΠ³ΠΎ Π½Π΅ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π²ΠΈΡΡΠ³ΡΠΈ ΠΏΡΠΎΡΡΠΎ ΡΠ°ΠΊ ΡΠΎΠ·ΡΠΌΡΡΡΡ ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΠΈ Π·Π°Π½Π°Π΄ΡΠΎ Π²ΡΡΡΠΊΠΎΠ° ΡΠΎΠ·ΡΠΈΡΡΡΠΈ Π΄ΡΡΡ Π½Π΅ ΠΌΠΎΠΆΠ½Π° Π²ΠΎΠ½Π° ΠΏΡΠΎΡΡΠΎ Π·Π°Π²Π°Π»ΠΈΡΡΡ ΡΠΎΠΌΡ Π²ΠΎΠ½ΠΈ ΡΡΠΈ Π΄ΠΎ Π±ΠΎΡ ΡΠΎΠ·ΠΊΠΎΠΏΡΡΡΡ Π°ΠΌΡΠ½Π΄Π°Π»ΡΠΊ Ρ ΠΏΠΎΠΊΠΈ ΠΏΡΠ°ΡΡΡ ΡΠ΅Ρ
Π½ΡΠΊΠΈ
|
76 |
-
```
|
77 |
-
|
78 |
-
long_1.wav (with News LM):
|
79 |
-
|
80 |
-
```
|
81 |
-
ΡΠ΅ΡΡΠ΅ ΡΠΈ Π΄ΠΈΠ²ΠΎΠ²ΠΈΠΆΠ½ΠΈΡ
ΠΏΠΎΡΡΡΡΠ½ΠΎΠΊ ΠΌΡΠ»ΡΠΉΠΎΠ½ΠΈ Π»ΡΠ΄Π΅ΠΉ ΡΠ°ΠΊΡΠΈΡΠ½ΠΎ Π² ΠΏΡΡΠΌΠΎΠΌΡ Π΅ΡΡΡΡ Π²ΠΆΠ΅ ΡΡΠΈ Π΄ΠΎΠ±ΠΈ ΡΠΏΠΎΡΡΠ΅ΡΡΠ³Π°ΡΡΡ Π·Π° ΡΠΏΡΠΎΠ±Π°ΠΌΠΈ ΠΌΠ΅ΡΠΎΠΊΠ°Π½ΡΡΠΊΠΈΡ
ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΡΠ² Π΄ΡΡΡΡΠΈΡΠΊΠΎΠ»ΠΎΠ΄Ρ Π·Π° ΠΏΡΡΠΈΡοΏ½οΏ½ΡΠ½Π΅ Ρ
Π»ΠΎΠΏΡ Π΄ΠΎΡΡ Π½Π΅Π·ΡΠΎΠ·ΡΠΌΡΠ»ΠΎ ΡΠΈ Π²Π΄Π°ΡΡΡΡ Π΄ΡΡΡΠ°ΡΠΈ ΠΉΠΎΠ³ΠΎ Π· ΡΡΠΈΠ΄ΡΡΡΠΈΠΌΠ΅ΡΡΠΎΠ²ΠΎΠ³ΠΎ ΠΏΡΠΎΠ²Π°Π»Π»Ρ ΠΆΠΈΠ²ΠΈΠΌ ΠΏΡΠΎ Π½Π°Π΄Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎ ΡΠΊΠ»Π°Π΄Π½Ρ ΠΎΠΏΠ΅ΡΠ°ΡΡΡ ΡΠΎ ΡΡΠΈΠ²Π°Ρ Π² ΡΡ ΠΌΠΈΡΡ Ρ Π½Π΅ ΡΠ»ΡΡΠ°ΡΡΡΠΊΠΎΠ΄ΡΠ·Π² ΡΠΊΠΈΠΉ ΠΏΡΠΎΠ²Π°Π»ΠΈΠ²ΡΡ ΠΏΡΡΠΈΡΡΡΠ½ΠΈΠΉ ΡΠ°ΡΠ½ Π»Π΅Π΄Ρ ΠΏΠΎΠΌΡΡΠ½Π° Π΄ΡΡΠ° Π² Π·Π΅ΠΌΠ»Ρ ΠΌΠ΅Π½ΡΠ΅ ΡΡΠΈΠ΄ΡΡΡΠΈ ΡΠ°Π½ΡΠΈΠΌΠ΅ΡΡΡΠ² Ρ Π΄ΡΠ°ΠΌΠ΅ΡΡΡ Π°Π»Π΅ Π² Π³Π»ΠΈΠ± Π²ΠΎΠ½Π° ΡΡΠ³Π½Π΅ΡΡΡΡ Π½Π° ΡΡΠΈΠ΄ΡΡΡΡ Π΄Π²Π° ΠΌΠ΅ΡΡΠΈ Π±Π°ΡΡΠΊΠΈ ΡΡΠΊΠ°Π»ΠΈ ΡΠΈΠ½Π° ΠΊΡΠ»ΡΠΊΠ° Π³ΠΎΠ΄ΠΈΠ½ ΠΏΠ΅ΡΠ΅Π΄ ΡΠΈΠΌ ΡΠΊ Π·ΡΠΎΠ·ΡΠΌΡΠ»ΠΈ Π²ΡΠ½ ΠΏΡΠ΄ Π·Π΅ΠΌΠ»Π΅Ρ Π° ΠΊΠΎΠ»ΠΈ Π²ΡΠ½ Π·Π½ΠΈΠΊ Ρ ΠΌΠΎΠ»ΠΈΠ»Π°ΡΡ Π±ΠΎΠ³Ρ ΠΏΡΠΎΡΠΈΠ»Π° Π°Π±ΠΈ Π°Π»Π°Π³Π±ΠΈΡΡΠ³ΠΌΠΎ ΡΠΈΠ½Π° Ρ ΠΉΠΎΠ³ΠΎ Π΄ΡΡΡΠ°Π»ΠΈ ΡΠΊΠΎΠ»ΠΎΡΡΠ·ΡΠΆΠΈΠ²ΠΈΠΌ Π³ΠΎΡΠΏΠΎΠ΄ΠΈ Ρ
Π°ΠΉ ΠΉΠΎΠΌΡ ΡΠ°ΠΌ ΠΌΠ΅Π½ΡΠ΅ Π±ΠΎΠ»ΠΈΡΡ Π² ΡΡΠΉ Π΄ΡΡΡ Ρ ΡΠ°ΠΊ ΡΠΏΠΎΠ΄ΡΠ²Π°ΡΡΠΈΡΡ ΡΠΎ Ρ ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΠΈ Π²ΡΠ΅ Π²ΠΈΠΉΠ΄Π΅ ΠΉΠΎΠ³ΠΎ Π½Π΅ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π²ΠΈΡΡΠ³ΡΠΈ ΠΏΡΠΎΡΡΠΎ ΡΠ°ΠΊ ΡΠΎΠ·ΡΠΌΡΡΡΡ ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΡΠ² Π·Π°Π½Π°Π΄ΡΠΎ Π²ΡΠ·ΡΠΊΠΎ Π° ΡΠΎΠ·ΡΠΈΡΡΡΠΈ Π΄ΡΡΡ Π½Π΅ ΠΌΠΎΠΆΠ½Π° Π²ΠΎΠ½Π° ΠΏΡΠΎΡΡΠΎ Π·Π°Π²Π°Π»ΠΈΡΡΡΡ ΡΠΎΠΌΡ Π²ΠΎΠ½ΠΈ ΡΡΠΈ Π΄ΠΎΠ±ΠΎΡ ΡΠΎΠ·ΠΊΠΎΠΏΡΡΡΡ Π°ΠΌΠ½Ρ Π΄Π°Π»ΡΠΊΡ ΠΏΠΎΠΊΠΈ ΠΏΡΠ°ΡΡΡ ΡΠ΅Ρ
Π½ΡΠΊ
|
82 |
-
```
|
83 |
-
|
84 |
-
long_1.wav (with better News LM):
|
85 |
-
|
86 |
-
```
|
87 |
-
ΡΠ΅ΡΡΠ΅ ΡΠΈ Π΄ΠΈΠ²ΠΎΠ²ΠΈΠΆΠ½ΠΈΠΉ ΠΏΠΎΡΡΡΡΠ½ΠΎΠΊ ΠΌΡΠ»ΡΠΉΠΎΠ½ΠΈ Π»ΡΠ΄Π΅ΠΉ ΡΠ°ΠΊΡΠΈΡΠ½ΠΎ Π² ΠΏΡΡΠΌΠΎΠΌΡ Π΅ΡΡΡΡ Π²ΠΆΠ΅ ΡΡΠΈ Π΄ΠΎΠ±ΠΈ ΡΠΏΠΎΡΡΠ΅ΡΡΠ³Π°ΡΡΡ Π·Π° ΡΠΏΡΠΎΠ±Π°ΠΌΠΈ ΠΌΠ°ΡΠΎΠΊΠ°Π½ΡΡΠΊΠΈΡ
ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΡΠ² Π΄ΡΡΡΡΠΈΡΠΊΠΎΠ»ΠΎΠ΄Ρ Π·Π° ΠΏΡΡΠΈΡΡΡΠ½Π΅ Ρ
Π»ΠΎΠΏΡ Π΄ΠΎΡΡ Π½Π΅Π·ΡΠΎΠ·ΡΠΌΡΠ»ΠΎ ΡΠΈ Π²Π΄Π°ΡΡΡΡ Π΄ΡΡΡΠ°ΡΠΈ ΠΉΠΎΠ³ΠΎ Π· ΡΡΠΈΠ΄ΡΡΡΠΈΠΌΠ΅ΡΡΠΎΠ²ΠΎΠ³ΠΎ ΠΏΡΠΎΠ²Π°Π»Π»Ρ ΠΆΠΈΠ²ΠΈΠΌ ΠΏΡΠΎ Π½Π°Π΄Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎ ΡΠΊΠ»Π°Π΄Π½Ρ ΠΎΠΏΠ΅ΡΠ°ΡΡΡ ΡΠΎ ΡΡΠΈΠ²Π°Ρ Π² ΡΡ ΠΌΠΈΡΡ Ρ Π½Π΅ ΡΠ»Π΅ΡΠ°ΡΡΡΠΊΠΎΠ΄ΡΠ·Π² ΡΠΊΠΈΠΉ ΠΏΡΠΎΠ²Π°Π»ΠΈΠ²ΡΡ ΠΏΡΡΠΈΡΡΡΠ½ΠΈΠΉ ΡΠ°ΡΠ½ Π»Π΅Π΄Ρ ΠΏΠΎΠΌΡΡΠ½Π° Π΄ΡΡΠ° Π² Π·Π΅ΠΌΠ»Ρ ΠΌΠ΅Π½ΡΠ΅ ΡΡΠΈΠ΄ΡΡΡΠΈ ΡΠ°Π½ΡΠΈΠΌΠ΅ΡΡΡΠ² Ρ Π΄ΡΠ°ΠΌΠ΅ΡΡΡ Π°Π»Π΅ Π² Π³Π»ΠΈΠ± Π²ΠΎΠ½Π° ΡΡΠ³Π½Π΅ΡΡΡΡ Π½Π° ΡΡΠΈΠ΄ΡΡΡΡ Π΄Π²Π° ΠΌΠ΅ΡΡΠΈ Π±Π°ΡΡΠΊΠΈ ΡΡΠΊΠ°Π»ΠΈ ΡΠΈΠ½Π° ΠΊΡΠ»ΡΠΊΠ° Π³ΠΎΠ΄ΠΈΠ½ ΠΏΠ΅ΡΠ΅Π΄ ΡΠΈΠΌ ΡΠΊ Π·ΡΠΎΠ·ΡΠΌΡΠ»ΠΈ Π²ΡΠ½ ΠΏΡΠ΄ Π·Π΅ΠΌΠ»Π΅Ρ Π° ΠΊΠΎΠ»ΠΈ Π²ΡΠ½ Π·Π½ΠΈΠΊ Ρ ΠΌΠΎΠ»ΠΈΠ»Π°ΡΡ Π±ΠΎΠ³Ρ ΠΏΡΠΎΡΠΈΠ»Π° Π°Π±ΠΈ Π°Π»Π°ΠΊΡΠ±ΠΈΡΡΠ³ΠΌΠΎ ΡΠΈΠ½Π° Ρ ΠΉΠΎΠ³ΠΎ Π΄ΡΡΡΠ°Π»ΠΈ ΡΠΊΠΎΠ²ΠΎΠ΄ΡΠ·ΡΠΆΠΈΠ²ΠΈΠΌ Π³ΠΎΡΠΏΠΎΠ΄ΠΈ Ρ
Π°ΠΉ ΠΉΠΎΠΌΡ ΡΠ°ΠΌ ΠΌΠ΅Π½ΡΠ΅ Π±ΠΎΠ»ΠΈΡΡ Π² ΡΡΠΉ ΡΡΡΡΡΡΠ°ΠΊ ΡΠΏΠΎΠ΄ΡΠ²Π°ΡΡΠΈΡΡ ΡΠΎ Ρ ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΠΈ Π²ΡΠ΅ Π²ΠΈΠΉΠ΄Π΅ ΠΉΠΎΠ³ΠΎ Π½Π΅ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π²ΠΈΡΡΠ³ΡΠΈ ΠΏΡΠΎΡΡΠΎ ΡΠ°ΠΊ ΡΠΎΠ·ΡΠΌΡΡΡΡ ΡΡΡΡΠ²Π°Π»ΡΠ½ΠΈΠΊΡΠ² Π·Π°Π½Π°Π΄ΡΠΎ Π²ΡΠ·ΡΠΊΠΎ Π° ΡΠΎΠ·ΡΠΈΡΡΡΠΈ Π΄ΡΡΡ Π½Π΅ ΠΌΠΎΠΆΠ½Π° Π²ΠΎΠ½Π° ΠΏΡΠΎΡΡΠΎ Π·Π°Π²Π°Π»ΠΈΡΡΡΡ ΡΠΎΠΌΡ Π²ΠΎΠ½ΠΈ ΡΡΠΈ Π΄ΠΎΠ±ΠΎΡ ΡΠΎΠ·ΠΊΠΎΠΏΡΡΡΡ Π°ΠΌΠ½ΠΎΠΊΠ΄Π°Π»ΡΠΊ Ρ ΠΏΠΎΠΊΠΈ ΠΏΡΠ°ΡΡΡ ΡΠ΅Ρ
Π½ΡΠΊ
|
88 |
-
```
|
89 |
-
|
90 |
-
tsn.wav (with better News LM):
|
91 |
-
|
92 |
-
```
|
93 |
-
ΡΠ½ΡΠΎΡΠΌΠ°ΡΡΠΉΠ½ΠΈΠΉ Π²Π΅ΡΡΡ Π½Π° ΠΎΠ΄ΠΈΠ½ ΠΏΠ»ΡΡ ΠΎΠ΄ΠΈΠ½ Π² ΡΡΡΠ΄ΡΡ ΡΠ° ΡΠ΅ΠΌ Π»ΡΠ΄ΡΡ ΡΠ°ΡΠ°Π½ Π° ΠΏΠ»Π°ΡΠΎΠ½ΠΎΠ²Π° ΠΊΠΎΡΠΎΠ»Π΅Π²Π° Π· ΡΠΎΠ³ΠΎ ΠΏΠΎΡΠΈΠ½Π°Π»ΠΎΡΡ Ρ ΡΠΈΠΌ ΡΠ΅ Π·ΠΎΠ²ΡΡΠΌ Π½Π΅ Π·Π°ΠΊΡΠ½ΡΠΈΡΡΡΡ ΠΏΡΠ°Π²Π»ΡΠ½Π½Ρ ΡΠ»ΠΈΠ·Π°Π²Π΅ΡΠΈ Π΄ΡΡΠ³ΠΎΡ ΡΡΠΌΠ΄Π΅ΡΡΡ ΡΠΎΠΊΡΠ² Π½Π° ΡΡΠΎΠ½Ρ ΠΎΠΊΠΌΠΎΠ²ΡΠ²ΡΠΏΠΎΠ»ΠΎΠ²Π½Ρ ΡΡΡΠ΅Π½Π½Ρ ΡΠΎΠ΄ΠΎ ΡΠΏΠ°Π΄ΠΊΠΎΡΠΌΡΡΠ² ΠΊΠΎΡΠΎΠ½ΠΈ Π±ΡΠ»ΡΡΠ΅ Π·Π°ΡΠ°Π· ΡΡΠ΅
|
94 |
-
```
|
95 |
-
|
96 |
-
tsn_2.wav (with better News LM):
|
97 |
-
|
98 |
-
```
|
99 |
-
Π΄ΠΎ ΠΎΡΠ»ΠΎ Π·ΡΠ»ΠΈ ΠΏΡΠ΄ ΡΠ°Ρ ΡΡΡΡΠ»ΡΠ½ΠΈΠ½ΠΈ Π½Π° ΠΏΡΠ²Π΄Π΅Π½ΠΌΠ°ΡΡ Π² ΠΆΠ°Π½Ρ Π²Π»ΡΡΠΈΠ»ΠΈ ΡΡΠΈ ΠΊΡΠ»Ρ ΠΎΠ΄Π½Π° Π· Π½ΠΈΡ
ΡΠΎΠ·ΡΡΠ²Π°Π»Π° ΠΊΠΈΡΠΊΡΠ²Π½ΠΈΠΊ ΡΠ° Ρ
ΡΠ΅Π±Π΅Ρ ΠΆΠΈΠ²ΠΎΡΡ ΠΊΠ°ΡΠ°ΡΡΡΠΎΡΡ ΡΠΎΠ·ΡΠΌΡΡΠΈ ΡΡΠΈ Π»ΡΡΡΠΈ ΠΊΡΠΎΠ²Ρ ΡΠΊΡ Π½Π΅ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ ΠΏΠΎΠ²Π΅ΡΠ½ΡΡΠΈ ΡΠΎΠΌΡ ΡΠΎ Π²ΠΎΠ½Π° Π·Π°Π±ΡΡΠ΄Π½Π΅Π½Π° ΠΊΠΈΡΠΊΡΠ²Π½ΠΈΠΊ ΠΊΠΈΠ½ΡΠ²ΡΠΈ Π΄ΠΎΠ· ΡΠΎΠ·ΡΡΠ²Π°Π½ΠΈΠΉ ΠΎΡΠ²ΡΡΡΠ½Π°Π²ΡΡΡ Π½Π΅ ΠΊΠΎΡΠΈΡΠ΅ΠΊΠ°ΠΏΠΎ ΡΠΆΠΈΠ»Π°ΡΡ Π΄ΡΠΆΠ΅ Π°ΠΊΡΠΈΠ²Π½Π° ΠΏΡΠΎΡΠΈ Π΄ΡΡΡΠ°Π»Π° ΠΊΡΠ»Ρ ΠΏΡΠΎΡΡΠΎ Π½Π° Π²Π°ΡΡΡ ΠΊΠΎΠ½ΡΡΠ°ΠΊΡΠ½ΠΈΡΡ Π²ΡΠ΄ΠΏΠΎΠ²ΡΠ΄Π°Π»Π° Π·Π° ΠΎΠΏΠΎΠ²ΡΡΠ΅Π½Π½Ρ ΡΠ°ΡΡΠΈΠ½ΠΈ ΠΏΡΠΎ Π½Π°Π΄Π·Π²ΠΈΡΠ°ΠΉΠ½Ρ ΡΠΈΡΡΠ°ΡΡΡ Π²ΡΠ½ Π±ΡΠ² Π² Π΄Π²ΠΎΡ
ΠΊΡΠΎΠΊΠ°Ρ
Π²ΡΠ΄ Π²Π΅Π»ΡΠΌΠΈ ΠΌΠΎΠ²ΡΠΊΠΈ ΡΠΎΠ·ΡΡΡΡΠ»ΡΠ»ΠΈ ΡΠΎ Π½Π° ΠΏΠΎΡΡΡ Π±ΡΠ»Π° ΡΠ°Π΄ΡΠΎΡΡΠ°Π½ΡΡΡ Ρ Ρ Π·Π°ΠΏΡΠ°Π· ΡΠ° ΠΌΠΎΠ³Π»Π° Π΄ΠΎΠΏΠΎΠ²ΡΡΡΠΈ ΡΠ°ΡΡΠΈΠ½Ρ Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ Π½Ρ Ρ Π½Π΅ Π·ΠΌΠΎΠ³Π»Π° Π΄ΠΎΠΏΠΎΠ²ΡΡΡΠΈ Π·ΡΠΎΠ±ΠΈΠ² ΠΏΠΎΡΡΡΡΠ»ΠΈ Ρ Π²ΠΆΠ΅ Π½Π΅ Π·ΠΌΠΎΠ³Π»Π° ΠΏΡΠ΄Π²Π΅ΡΡΠΈΡΡ ΡΡΡΡΠ»ΡΠ½ΠΈΠ½Π° Π½Π° ΠΏΡΠ²Π΄Π΅Π½ΠΌΠ°ΡΡ ΡΡΠ°Π»Π°ΡΡ Π²Π½ΠΎΡΡ Π΄Π²Π°Π΄ΡΡΡΡ ΡΡΠΎΠΌΠΎΠ³ΠΎ ΡΡΡΠ½Ρ ΠΏΡΡΠ΅ΡΠΎ Π»ΡΠ΄Π΅ΠΉ Π·Π°Π³ΠΈΠ½ΡΠ»ΠΈ ΡΠ΅ΡΠ΅Π΄ Π½ΠΈΡ
ΠΎΠ΄Π½Π° ΡΠΈΠ²ΡΠ»ΡΠ½Π° ΡΠ΅ ΠΏΡΡΡΠΎ ΠΏΠΎΡΠ°Π½Π΅Π½Ρ ΠΏΡΠ΄ΠΎΠ·ΡΡΠ²Π°Π½ΠΈΠΉ Π΄Π²Π°Π΄ΡΡΡΠΈΡΡΡΠ½ΠΈΠΉ ΡΠΎΠΊΠΎΠ²ΠΈΡ
Π°ΡΡΠ΅ΠΌΡΡΡΠ²ΡΡΡ
Ρ ΡΡΠ·ΠΎ ΡΠΏΠΎΡΠ°ΡΠΊΡ ΠΏΡΠΎΠ²ΠΈΠ½Ρ Π²ΠΈΠ·Π½Π°Π²Π°Π² ΡΠ΅ΠΏΠ΅Ρ Ρ Π·Π½ΠΎΠ²Ρ ΠΌ Π°Π΄Π²ΠΎΠΊΠ°ΡΠ°ΠΌΠΈ Π²ΡΠ΄ ΠΏΠΎΠΊΠ°Π·ΡΠ² Π²ΡΠ΄ΠΌΠΎΠ²ΠΈΠ²ΡΡ Π°Π½Ρ ΠΊΠΎΠ½ΡΠ»ΡΠΊΡΠ½ΠΈΡ
ΡΠΈΡΡΠ°ΡΡΡ Π°Π½Ρ Π±ΠΎ ΡΠΎΠ³ΠΎ ΠΏΡΠ΄ΠΎΠ·ΡΡΠ»ΠΎΠ³ΠΎ ΠΆΠ°Π½Π½Π° ΠΊΠ°ΠΆΠ΅ Π΄ΠΎ ΡΡΡΡΠ»ΡΠ½ΠΈΠ½ΠΈ ΡΠ· ΠΏΡΠ΄ΠΎΠ·ΡΡΠ²Π°Π½ΠΈΠΌ Π½Π°Π²ΡΡΡ Π½Π΅ ΡΠΎΠ·ΠΌΠΎΠ²Π»ΡΠ»Π° Π° ΡΡΠ»ΡΠΊΠΈ Π·Π° ΠΏΠ°ΠΌΡΡΠ°Π»Π° Ρ ΠΎΡΠ΅ Π° ΡΠ°ΠΊ Ρ Π½Π°Π²ΡΡΡ ΠΉΠΎΠ³ΠΎ ΡΠΌΠ΅Π½Ρ ΠΉ ΡΠ°ΠΌΡΠ»ΡΡ Π½Π΅ Π·Π½Π°Π² Ρ Π΄ΡΠΆΠ΅ ΡΡΠ΄ΠΊΠΎ Π·Π°ΡΡΡΠΏΠ°Π»Π° ΡΡ ΠΊΠ°ΡΡ ΡΠ°ΠΌ Π΄Π΅ ΡΡΠΎΡΠ½ΠΎΠ³ΠΎ ΡΠ»ΡΠΆΠ±Π° Ρ Π½Π΅ ΠΌΠΎΠΆΡ Π·Π½Π°ΡΠΈ ΡΠΎΠΌΡ Π²ΡΠ½ ΡΠ°ΠΊ Π·ΡΠΎΠ±ΠΈΠ² Π½Π°Π²ΠΏΡΠΎΡΠΈ ΠΆΠ°Π½ΠΈ ΡΠ΅Π°Π½ΡΠΌΠ°ΡΡΡ ΠΊΠΎΠ½ΡΡΠ°ΠΊΡΠ½ΠΈΠΊ ΡΠ³ΠΎΡ ΡΠ°ΠΌΠ΅Π½Π΅Π½ΠΊΠΎ ΠΏΡΠ΄ ΡΠ°Ρ ΡΡΡΠ»ΡΠ½ΠΈΠ½ΠΈ Π²ΡΠ΄ΠΏΠΎΡΠΈΠ²Π°Π² Π·Π° Π³ΡΠ°ΡΡΠΊΠΎΠΌ ΠΏΠΎΠ²ΠΎΡΠΎΠΆΠ΅ Π·Π°Π±ΡΠ³Π°Ρ ΡΠΈΠΊΡΡΠΈΡΠΎΡΠ°ΠΌΡΡΡ ΠΊΡΡΡ ΡΡΠΌΠ½ΠΎΡ Π½Π°ΡΠ΅
|
100 |
-
```
|
101 |
-
|
102 |
-
mer_lviv_interview.wav:
|
103 |
-
|
104 |
-
```
|
105 |
-
ΠΌΠΈ ΠΎΡΡΠΈΠΌΠ°Π»ΠΈ Π½ΠΎΠ²ΠΎΠ³ΠΎ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠ° ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΠΎΠ±Π»Π°ΡΡΡ Π° ΡΠ΅ Ρ ΡΠ°Ρ
ΠΎΠ²Π° Π»ΡΠ΄ΠΈΠ½Π° Ρ Π· Π΄ΡΡΠ½Π΅ ΠΌΡΡΡΠ° ΠΌΠΈ ΠΏΠ΅ΡΠ΅ΡΠΎΡΠΌΠΎΡΡΠ²Π°ΠΈ Π½Π°Ρ ΡΠΏΡΠ°Π²Π»ΡΠ½Π½Ρ Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΈΡ
ΡΠΈΡΡΠ°ΡΡ ΡΡΠΎΠ³ΠΎΠ΄Π½Ρ Π²ΠΎΠ½ΠΎ ΠΌΠ°Ρ ΡΡΠ½ΠΊΡΡΡ Π½Π°Π·Π²ΠΈΡΠΉΠ½Ρ
ΡΠΈΡΡΠ°ΡΡΠ²ΡΠ»ΡΠ½ΠΎΠ³ΠΎ Π·Π°Ρ
ΠΈΡΡ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Ρ Π² Π½Π°Ρ Π΄ΠΎΡΡΠ°ΡΠ½ΡΠΎ ΡΠΊΡΡΠ½Π° ΡΠΏΡΠ²ΠΏΡΠ°Ρ ΡΠΊΡΡΠ½Π° ΠΊΠΎΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ ΡΠΎΠΌΡ ΡΠΎ Π·Π°Π½Π°ΡΠ°ΠΌΡΡΡΠ° Π΄ΠΎΠΏΠΎΠΌΠΎΠ³ΡΠΈ Π΄Π΅ΡΠΆΠ°Π²Π½ΠΎΠΌΡ ΡΠ°ΠΊΡΠΈΡΠ½ΠΎ ΠΊΡΡΡΠ½ΠΊΡ ΡΠΎΠ± Π²ΡΠ½ ΡΠΊΡΡΠ½ΠΎ Π·Π° ΠΏΡΠ°ΡΡΠ²Π°Π² ΡΡΠΎΡΠΎΠ²Π½ΠΎ ΡΠ΅ΠΌΠΎΠ½ΡΠΈ ΡΡΠΎΡΠΎΠ²Π½ΠΎ ΠΎΠ±Π»Π°Π΄Π½Π°Π½Π½Ρ Ρ ΡΡΠΎΡΠΎΠ²Π½ΠΎ ΡΠ΅ΠΊΡΡΡΠ΅Π½Π³Ρ Π»ΡΠ΄Π΅ΠΉ ΡΠΊΡ Π·Π°ΡΠ°Π· Π³ΠΎΠ»ΠΎΡΡΠ»ΠΈΡΡ ΡΠΎ Π±ΠΈ Π±ΡΡΠΈ ΠΏΠΎ ΠΊΠΎΠ½ΡΡΠ°ΠΊΡΠ°Ρ
Π² ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½Ρ Π½Π° ΡΠ΅ΡΠΈΡΠΎΡΡΡ ΠΌΡΡΡΠ° Π»ΡΠ±ΠΎΠ²Π° Π½Π° ΡΡΠ²ΡΡΡΡΡΡΡ Π΄Π²Π±Π°ΡΠ°ΡΠΎΠ½ΠΈ Ρ Π½Π°ΡΠ°Π·Ρ Π²ΡΠ΅ ΠΉΠ΄Π΅ ΡΡΡΠΊΠΎ Π·Π° ΡΠΈΠΌΠ»Π°Π½Π°ΠΌ ΡΠΊΡ Π²ΠΈΠ·Π½Π°ΡΠ΅Π½ΠΎ Π΄Π΅ΡΠΆΠ°Π²ΠΎΡ Π½Π°ΡΠΎΠΌΡΡΡΡ ΠΌΡΡΡΠΎ Π²Π·ΡΠ»Π° Π½Π° ΡΠ΅Π±Π΅ Π΄ΠΎΠ΄Π°ΡΠΊΠΎΠ²ΠΎ ΡΡΠ½ΠΊΡΡΡ ΠΌΠΈ ΠΎΡΠ°Π»ΠΈ Π²Π΅Π»ΠΈΠΊΠΈΠΉ Π½Π°Π²ΡΠ°Π»ΡΠ½ΠΈΠΉ ΠΏΡΠΎΡΠ΅Ρ Π΄Π»Ρ ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΡΠ² ΠΊΠΎΠΌΡΠ½Π°Π»ΡΠ½ΠΈΡ
ΠΏΡΠ΄ΠΏΡΠΈΡΠΌΡΠΎΡΠΎΠ±Π»ΠΈΠ²ΠΎ ΠΌΠΎΠ² Π΄Π»Ρ ΠΏΡΠΎ ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½Π΅ ΠΏΡΠ΄ΠΏΡΠΈΡΠΌΡΡΠ²Π° ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΠΈ Π²ΠΈΠΊΠΎΠ½Π°Π²ΡΠΈΠΉ Ρ
ΠΎΡΠΎΠ½Π½Ρ Π²Π»Π°Π΄ΠΈ Π±Π°Π½Π°Π»ΡΠ½ΠΎ Π²ΡΠ΄Π½ΠΎΠ²ΠΈΡΠΈ Π½Π°Π²ΠΈΠΊΠΈ Π²ΠΎΠ»ΠΎΠ΄ΡΠ½Π½Ρ Π·Π±ΡΠΎΡ Π² Π½Π°ΡΠΈΡ
ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΡΠ² Π»ΡΡΠ²Π°ΠΆΠ»ΠΈΠ²ΠΈ ΡΠΈΠ½ΠΈΠΊΡΠ΅ ΡΠΌΡΠ½Π½Ρ Π½Π°Π΄Π°Π²Π°ΡΠΈ ΠΏΠ΅ΡΡΡ Π΄ΠΎ ΠΌΠ΅Π΄ΠΈΡΠ½Ρ Π΄ΠΎΠΏΠΎΠΌΠΎΠ³Ρ Ρ Π·Π° ΡΠ°ΠΊΠ° Π²Π΅Π»ΠΈΠΊΠ° Ρ
Π²ΠΈΠ»Ρ ΡΠΊΠ° ΠΏΡΡΠ»Ρ Π·Π°ΠΊΡΠ½ΡΠ΅Π½Π½Ρ ΠΊΠ°ΡΠ°Π½ΡΠΈΠ²Π½ΠΎ Π΄Π΅ Π²ΠΊΠ»ΡΡΠ°ΡΠΈ Ρ ΡΡΠ°ΡΡΠΎΠΊΠ»Π°ΡΠ½ΠΈΠΊΡΠ² ΠΌΠΈ Π½Π° ΠΏΡΠΈ Π²Π΅Π»ΠΈΠΊΠΈΠΉ ΠΆΠ°Π»Ρ Π±Π°Π³Π°ΡΠΎ Π² ΠΎΡΡΠ½Π½ΡΡ
ΡΠΎΠΊΡΠ² ΡΠ°ΠΊΠ΅ Π²ΡΠ°ΠΆΠ΅Π½Π½Ρ ΡΠΎ ΠΏΠ΅ΡΠ΅Π±ΡΠ²Π°Π»ΠΈ Π»ΡΡΡΠΈΠ½ΠΎΠΌΡ ΡΠ½Ρ ΡΠΎΠ·ΡΠΌΡΡΡΠΈ ΡΠΎ ΡΠΎΡΡΡ ΠΏΠΎΡΡΡ Ρ Π·Π° Π²ΠΈ Π±ΡΠ΄Π΅ ΡΠ΅Π½ΡΡ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π°Π³ΡΠ΅ΡΡΡ Π½Π°ΠΌ ΡΡΠ΅Π±Π° Π²ΠΈΡΠΊΠΎΠ»ΡΠ²Π°ΡΠΈ ΡΠ΅Π±Π΅ Ρ Π±ΡΡΠΈ Π³ΠΎΡΠΎΠ²ΠΈΠΌΠΈ Π΄ΠΎ ΡΠΎΠ³ΠΎ ΡΠΎΠ± Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ ΡΡΠ΄Π½Ρ Π΄Π΅ΡΠΆΠ°Π²Ρ ΡΠΊΡΠΎ ΠΊΠΎΠΆΠ΅Π½ ΡΠΊΡΠ°ΡΠ½Π΅ΡΡ Π±ΡΠ΄Π΅ Π΄ΠΎΠ±ΡΠ΅ Π²ΠΎΠ»ΠΎΠ΄ΡΡΠΈ Π½Π°Π²ΠΈΠΊΠ°ΠΌΠΈ ΡΡΡΠΈΠ»Π΅ΡΡΠΊΠΎΡ Π·Π±ΡΠΎΡ Π±ΡΠ΄Π΅ Π²ΠΌΡΡΠΈ Π½Π°Π΄Π°Π²Π°ΡΠΈ ΠΌΠ΄ΠΈΡΠ½Ρ Π΄ΠΎΠΏΠΎΠΌΠΎΠ³Ρ Π² Π±ΠΈΠΌΠ΅Π½Π΅ ΡΠ΅ ΠΏΠΎΡΡΠΆΠ½Π° ΡΠΈΠ»Π° ΡΠΊΡ Π½Π΅ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π·Π΄ΠΎΠ»Π°ΡΠΈ ΡΠΎΠΌΡ Π² ΠΊΠΎΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ Π· Π΄Π΅ΡΠΆΠ°Π²Ρ Ρ ΡΡΡ
Π°ΡΠΌΠΎΡΡ Π²ΠΏΠ΅ΡΠ΅Π΄ ΡΠ΅ ΡΠ°ΠΊΠ΅ ΠΏΠΈΡΠ°Π½Π½Ρ ΠΎΡΡ Π½Π°Ρ Π±Π°Π³Π°ΡΠΎ ΠΊΠ°ΠΆΡΡΡ ΡΠΎ ΠΏΠΎΡΡΡ Ρ Π±ΡΠΈΠ³Π°Π΄Π° ΠΌΠΈ ΡΠΈΠ» ΡΠ΅ΡΡΡΠ°Π»ΡΠΎ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΠΌΠΎΠΆΡΡΡ ΡΡΠ²ΡΡΠ²Π°ΡΠΈΡΡ Ρ ΡΠΎΡΠΌΡΠ²Π°ΡΠΈΡΡ Π΄ΠΎΡΠΎΠ²ΠΎΠ»ΡΡΡ ΠΎΠ±ΡΠ΄Π½Π°Π½Π½Ρ ΡΠΈ Ρ Π²ΡΠ»Π²ΠΎΠ²Ρ ΡΠ΅ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ΅Π½ΠΎ Ρ ΡΠΊ Π²Π°ΡΠ΅ ΡΡΠ°Π²Π»Π΅Π½Π½Ρ Π΄ΠΎ ΡΠ°ΠΊΠΎΠ³ΠΎ Π²ΡΡΠ΅ ΠΌΠΈ ΠΏΡΠ°ΡΡΡΠΌΠΎ Π² ΡΡΡΠΊΡΠΉΠΊΠΎΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ Π· Π΄Π΅ΡΠΆΠ°Π²Π½ΠΈΠΌΠΈ ΡΠ½ΡΡΠΈΡΡΡΡΡΠΌΠΈ Ρ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½ΡΡΠ½ΡΠΉ Π΄Π΅Π½Ρ Ρ Π½Π΅ Π±Π°ΡΡ ΠΏΠΎΡΡΠ΅Π±ΠΈ Π² ΡΠ½ΡΠΈΡ
ΡΠΎΡΠΌΡΠ²Π°Π½Π½Ρ Ρ
ΡΠΎΠΌΡ ΡΠΎ ΠΏΡΠΈΠ½ΡΠΈΠΏΡ ΠΌΠ°Ρ Π±ΡΡΠΈ ΠΆΠΎΡΡΡΠΊΠ° Π²Π΅ΡΡΠΈΠΊΠ°Π»ΡΠ½ΠΎ ΠΌΠ°Ρ Π±ΡΡΠΈ ΠΎΠ΄ΠΈΠ½ ΡΠ΅Π½ΡΡ ΠΏΡΠ°Π²Π»ΡΠ½Π½Ρ ΡΠ»Π° ΠΎΠ±ΠΎΠ³ΠΎΡΠΎ Ρ Π½ΠΎΠ²ΠΈΠΉ ΠΊΠΎΠΌΠ°Π½Π΄ΡΠ²Π°Ρ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΡΠ°ΠΌΠΊΠ°Ρ
Π΄Π΅ΡΠΆΠ°Π²ΠΈ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊ ΡΡΠ°Π±ΠΎΠ½ΠΎΠ²ΠΈΠΉ ΠΏΡΠΈΠ½ΡΠΈΠΏΡ ΠΏΠ°Π½ Ρ ΡΠΎ ΡΠΊΠΈΠΉ Π² Π½Π°Ρ Ρ ΡΡΠ»ΠΈ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½Ρ ΠΎΡΠΎΠ½Ρ ΠΎΠ±Π»Π°ΡΡΡ Π½Π° ΡΠ΅ Π΄ΡΠΆΠ΅ ΡΠ°Ρ
ΠΎΠ²Π° ΠΊΠΎΠΌΠΏΠ΅ΡΠ΅Π½ΡΠ½Π° Π»ΡΠ΄ΠΈΠ½Π° Π· Π½ΠΈΠΌ ΠΏΡΠΎΡΡΠΎ ΡΡΠΌΠ½ΠΎ ΡΡΠ² ΠΏΡΠ°ΡΡΠ²Π°ΡΠΈ Π² Π½Π΅ ΡΠΎΠ·ΡΠΌΡΡ Π²ΡΠ΅ ΡΠΏΡΠ²ΡΠ»ΠΎΠ² Π·Π°ΠΊΠΎΠ½ΠΎΠΌ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ΅Π½Π° ΠΌΠΎΠΆΠ»ΠΈΠ²ΡΡΡΡ ΠΏΡΠΈΠΊΠΈΠ΄Π°Π½Π½Ρ Π² ΡΠ°Π·Ρ ΡΠ°ΠΊΠΎΡ Π½Π°Π³Π°Π»ΡΠ½ΠΎΡ ΠΏΠΎΡΡΠ΅Π±ΠΈ ΠΏΠ΅ΡΠ΅ΠΊΠΈΠ΄Π°Π½Π½Ρ Π±ΡΠΈΠ³Π°Π΄Π° ΡΠΈΠ» ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π² Π·ΠΎΠ½ΠΎΠ±ΠΎΡΠ²ΠΈΡ
Π΄Ρ Π²ΡΡΡ ΠΎΠ±Π»Π°ΡΡΡ ΡΠΊΠ΅ Π²Π°ΡΠ΅ ΡΡΠ°Π²Π»Π΅Π½Π½Π΄ΠΎΡΡΠΎΠ³ΠΎ Ρ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π²Π°ΠΌ Π²ΡΠ΄ΠΎΠΌΠΎ ΡΠΊοΏ½οΏ½ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ ΡΡΠ°Π²Π»Π΅Π½Π½Ρ Ρ ΡΠ°ΠΌΠΈΡ
ΡΡΠ² ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π²ΡΡΡ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½ΡΡΠ½ΡΠΉ Π΄Π΅Π½Ρ ΠΌΠΈ ΡΠΎΠ·ΡΠΌΡΡΠΌΠΎ ΡΠΎ Π²ΡΠ½ ΡΠΎΡΡΡΠΉΠ½ΠΎΠΌΡ ΠΏΠ»Π°Π½Ρ ΡΡΠ΅Π±Π° Π·ΡΠΎΠ±ΠΈΡΠΈ ΡΠΈΡΠ°Π½ΡΡΠ½ΠΈΠΎΠ±ΡΠΌ ΠΏΡΠ°ΡΡ Π»ΡΠ΄ΡΠΌ ΡΡΠ΅Π±Π° ΠΏΠΎΡΡΠ½ΠΈΡΠΈ ΡΠΎ ΡΠΈ Π½Π΅ ΠΌΠΎΠΆΠ΅Ρ ΡΡ
ΠΎΠ²Π°ΡΠΈΡΡΡ Π²Π»Π°ΡΠ½ΠΎΠΌΡ ΠΏΠΎΠΌΠ΅ΡΠΊΠ°Π½Ρ ΠΊΠΎΠ»ΠΈ ΠΏΡΠΈΠΉΠ΄Π΅ Π²ΠΎΡΠΎΠ³ Π²ΠΎΡΠΎΠ³Π° ΡΡΠ΅Π±Π° Π½Π΅ ΠΏΡΡΡΠΈΡΠΈ Π½Π° ΡΠ΅ΡΠΈΡΠΎΡΡΡ Π½Π°ΡΠΎΡ ΠΊΡΠ°ΡΠ½ΠΈ Ρ Ρ Π΄ΡΠΌΠ°Ρ ΡΠΎ Π²ΡΡ Π³ΡΠΎΠΌΠ°Π΄ΡΠ½ΠΈ ΠΏΠΎΠ²ΠΈΠ½Π½Ρ ΠΌΠ°ΡΠΈ Π³ΠΎΡΠΎΠ²Π½ΡΠ·Π°Ρ
ΠΈ ΡΠ°ΡΠΈ ΡΠ²ΠΎΡ ΡΡΠ΄Π½Ρ ΠΊΡΡΠ½Ρ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½ΡΡΠ½ΡΠΉ Π΄Π΅Π½Ρ ΠΉΠ΄Π΅ Π½ΠΎΡΠΌΠ°Π»ΡΠ½ΠΈΠΉ ΠΏΡΠΎΡΠ΅Ρ Π»ΡΠ΄ΠΈ Π·Π°ΠΏΠΈΡΡΡΡΡΡΡ Π΄ΡΠΆΠ΅ Π±Π°Π³Π°ΡΠΎ Π²ΡΠ΄ΠΎΠΌΠ΅Π»ΡΠ΄Π΅ΠΉ Π·Π°ΠΏΠΈΡΡΡΡ Π² ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½Ρ ΠΎΠ±ΠΎΡΠΎΠ½Ρ Ρ Π²Π²Π°ΠΆΠ°Ρ ΡΠ΅ ΠΏΡΠ°Π²ΠΈΠ»ΡΠ½ΠΎ ΡΠΎΠΌΡ ΡΠΎ ΡΡΠ΅Π±Π° ΠΌΠ°ΡΠΈ ΡΠ°ΠΊΠΈ Π²ΠΈΡΡΠ² Π° ΡΡΠΎΡΠΎΠ²Π½ΠΎ Π²ΡΡΡ
ΡΠ½ΡΠΈΡ
ΡΠ΅ΡΠ΅ΠΉ ΡΠ΅ ΡΠ°Π· Π½Π°Π³ΠΎΠ»ΠΎΡΡΡ ΠΊΠΎΠ»ΠΈ ΠΌΠΎΠ²Π° ΠΉΠ΄Π΅ ΠΏΡΠΎ Π±Π΅Π·ΠΏΠ΅ΠΊΡ ΠΏΡΠΎ ΠΎΠ±ΠΎΡΠΎΠ½Ρ Π΄Π΅ΡΠΆΠ°Π²ΠΈ ΠΌΠ°Ρ Π±ΡΡΠΈ ΡΡΠ±ΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ Ρ Ρ ΡΡΡΠΊΠ° Π΄Π΅ΡΠΆΠ°Π²Π½Π° Π²Π΅ΡΡΠΈΠΊΠ°Π»ΡΡΡΠ΅Π±Π° Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Π±ΡΠ΄Π΅ΠΌΠΎ ΡΡ ΡΠ°Π·ΠΎΠΌ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Π½Π΅ΠΌΠ° Π½Π°ΡΠΎΡΠ°Π΄ΠΈΠ° ΡΠ΅ ΠΎΠ΄Π½Π΅ ΠΏΠΈΡΠ°Π½Π½Ρ Π±Π°Π³Π°ΡΠΎ Ρ
ΡΠΎ Π²ΠΏΠ΅ΠΎΠ½ΠΈΠΉΡΠΎ Π·Π°ΠΊΠΎΠ½ΡΠ½Π΅ Π΄ΠΎΡΡΠ°ΡΠ½ΡΠΎ Π²Π»Π°Π΄ΠΈ ΠΏΡΠΎ ΠΏΠΈΡΠ°Π½ΠΎ Π΄Π»Ρ ΠΌΡΡΡΠ΅Π²Π΅ Π°Π΄ΠΌΡΠ½ΡΡΡΡΠ°ΡΡΡ Π²Π»Π΄ΠΈΠΌΠ°ΡΡΡΡΡ Π½Π° ΡΠ²Π°Π·Ρ ΠΏΠΎΠ²Π½ΠΎΠ²Π°ΠΆΠ΅Π½Ρ Ρ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎΡΡΡ ΠΏΡΠΈ ΡΠΎΡΠΌΡΠ²Π°Π½Π½Ρ ΡΠΎΠ·Π³ΠΎΡΡΠ°Π½Π½Ρ ΡΠΈΠ» ΡΠΎΠ΅ΠΈΡΡΠ°Π»ΡΠ½ΠΎ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π²ΠΈ Π½Π° ΠΏΡΠ°ΠΊΡΠΈΡΡ Π·ΡΡΠΊΠ½ΡΠ»ΠΈΡΡ Π²ΠΆΠ΅ ΡΠΈΠΌ Ρ ΡΠΊ Π²ΠΈ ΠΎΡΡΠ½ΡΡΡΡ ΡΡ ΡΠ΅ Π½Ρ ΠΏΠΎ ΠΏΠ΅ΡΡΠ΅ Π΄ΠΎΠ±ΡΠ΅ ΡΠΎ Π·Π°ΠΊΠΎΠ½ Ρ Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎ ΡΠΎ ΠΎΠ΄ ΠΆΠΎΠ΄Π΅Π½ Π·Π°ΠΊΠΎΠ½ Π½Π΅ Ρ ΡΠ° ΠΎΠ±ΡΠ°Π»ΡΠ½ΠΈΠΉ ΡΠΈ ΡΠ΄Π΅Π°Π»ΡΠ½ΠΈΠΉ Π²ΡΠ½ ΠΌΠ°Π² Π±ΠΈ ΠΏΡΠΎΡΠ΅ΡΡ ΡΠ΅Π°Π»ΡΠ½ΠΎ ΠΆΠΈΡΡΡ Π·Π° Π·Π½Π°ΡΠΈ ΠΏΠ΅Π²Π½ΠΈΡ
ΠΊΡΡΠ³ΡΠ²Π°Π½Ρ Ρ Π½Π°ΠΏΡΠΈΠΊΠ»Π°Π΄ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠΎΠΌ ΡΠ΅ΡΠΈΡΡΠ°Π»ΡΠ½ΠΎΠΉ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π² ΡΠ°ΡΡ Π½Π° Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎΡ ΡΠΈΡΡΠ°ΡΡΡ Ρ Π³ΠΎΠ»ΠΎΠ²Π° ΠΎΠ±Π»Π°ΡΠ½ΠΎΡ Π°Π΄ΠΌΡΠ½ΡΡΡΡΠ°ΡΡΡ ΡΠ°ΠΊ Π° ΠΏΠΎ ΠΌΡΡΡΠΎ Π»ΡΠ±ΠΎΠ²ΠΎ Π½Π°ΠΏΡΠΈΠΊΠ»Π°Π΄ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠΎΠΌ Ρ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊ ΡΠ°ΠΉΠΎΠ½Π½ΠΎΡ Π°Π΄ΠΌΡΠ½ΡΡΡΡΠ°ΡΡΡ ΡΠ΅ Ρ Π½Π΅ Π²Π΅Π»ΠΈΠΊΠΈΠΉ ΠΎΡΡΡΡ Ρ Π΄Π΅ΠΊΡΠ»ΡΠΊΠ° ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΡΠ² Ρ Π²ΠΎΠ½ΠΈ ΠΌΠ°Π»ΠΈ Π±ΠΈ ΡΡΠΎΡΡΠΈ Π½Π°Π΄ Π³ΡΠΎΠΌΠ°Π΄ΠΈ ΠΌΡΡΡΠ° Π»ΡΠΎΠ° Ρ Π½Π΅ Π΄ΡΠΌΠ°Ρ ΡΠΎ ΡΠ΅ Π±ΡΠ΄Π΅ ΠΌΠ°ΡΠΈ Π΄ΠΎΡΡΠ°ΡΠ½Ρ ΡΠΊΡΡΠ½ΠΈΠΉ Π΅ΡΠ΅ΠΊΡ ΡΠΎΠΌΡ ΡΠΎ ΠΏΡΠΈΠ½ΡΠΈΠΏ Ρ ΠΊΡΡΠ²Π½ΠΈΠΊΠΎΠ±Π»Π°ΡΡΡ Ρ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊ ΠΌΡΡΠ° Π°Π»Π΅ ΡΡ ΡΠ΅ΡΡ Π½Π°ΠΏΠ΅Π²Π½ΠΎ ΠΊΠΎΠ»ΠΈ ΠΏΠΈΡΠ°Π»ΠΈ Π·Π°ΠΊΠΎΠ½ Π½Π΅ Π΄ΠΎ ΠΊΡΠ½ΡΡ ΠΏΡΡΠ°Ρ
ΠΎΠ²ΡΠ²Π°Π»ΠΈ Π°Π»Π΅ ΠΌΠΈ ΠΌΠ°ΡΠΌΠΎ ΡΠ°Ρ ΡΠΎΠ± ΡΠ΅ ΠΏΠΎΠΏΡΠ°Π²ΠΈΡΠΈ Π°Π»Π΅ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½Ρ ΠΌΠΈ ΡΡΡΠΊΠΎ Π²ΠΈΠΊΠΎΠ½ΠΈΠΎΠ·Π°ΠΊΠ½ΠΈ ΡΠΊΠΈΠΉ Ρ ΡΠ½ΡΠΈΡ
Π²Π°ΡΡΠ°Π½ΡΡΠ² Π½Π΅ ΠΌΠΎΠΆΠ΅ Π±ΡΡΠΈ ΠΏΠ°Π½Π΅ Π°Π½Π΄ΡΡΡΠ±ΡΠΊΠ²Π°Π»ΡΠ½ΠΎ ΠΎΡΡΠ°Π½Π½Ρ Π·Π°ΠΏΠΈΡΠ½Π½ΡΠ²Π°ΡΠ° Π±ΡΠΈΠ³Π°Π΄Π° ΡΠ»Π²ΡΠ²ΡΡΠΊΠ°Π½Ρ Π΄Π²Π° Π±Π°ΡΠ°Π»ΡΠΉΠΎΠ½ ΡΠΊ Π²ΠΈΠΊΡΠΈΡΠΈ ΡΠΊ Π²ΠΎΠ½ΠΈ Π·Π°ΡΠ°Π· ΠΎΠ·Π±ΡΠΎΡΠ½Ρ ΠΏΠΎΠ²Π½ΡΡΡΡ ΡΠΈ Π½ΡΡ ΡΠ΅ ΡΡΠ»ΡΠΊΠΈ ΡΡΡΠΈΠ»Π΅ΡΡΠΊΠ΅ Π·Π±ΡΡΡΠΈΡΠ΅ ΠΉ Π±ΡΠ»ΡΡ Π²Π°ΠΆΠΊΠ΅ ΠΎΠ·Π±ΡΠΎΡΠ½Π½ Ρ ΡΠΊ Π² ΡΡΠ°Π²ΠΈΡΠΈΡ Π΄ΠΎ ΡΠΎΠ³ΠΎ ΡΠΎΠ± ΠΌΡΡΡΠΊΠ΄Π½ΡΡΡΡΠ°ΡΡΡ ΠΌΠ°Π»Π° ΠΌΠΎΠΆΠ»ΠΈΠ²ΡΡΡΡΠΊΠ°ΠΆΡΠΌΠΎ ΡΠΎΡΡ ΠΊΡΠΏΡΠ²Π°ΡΠΈ Π΄Π»Ρ Π·Π°Π±Π·ΠΏΠ΅ΡΠ΅Π½Π½Ρ ΡΠ°ΠΌΠ΅ ΡΠΈ ΡΡΡΡΡΠ°Π»ΡΠΎΠΎΡΠΎΠ½ Ρ Π½Π° ΠΏΡΠΈΠΊΠ»Π°Π΄ ΡΡ ΠΆ ΡΠ°ΠΌΡ Π²ΠΈΡΠΏΡΠ»ΠΎΡΠ½ΠΈΠΊΠΈ Π΄Π»Ρ ΡΠΎΠ·Π²ΡΠ΄ΠΊΠΈ Π°Π±ΠΎ Π·Π°ΡΠΎΠ±ΠΈ Π·Π²ΡΠ·ΠΊΡ Π° Π°Π±ΠΎ ΠΎΡΡ ΡΠΎΡΡ ΡΠ°ΠΊΠ΅ Π½Ρ ΡΠΎ Π²ΠΈ ΡΠΎΠ·ΡΠΌΡΠ»ΠΈ ΠΌΠΈ Π΄ΠΎΡΡΠ°ΡΠ½ΡΠΎ Π±Π°Π³Π°ΡΠΎ ΠΏΠΎΠΌΠ°Π³Π°ΡΠΌΠΎ Π²ΡΡΠΌ Π½Π°ΡΠΈΠΌ Π²ΡΠΉΡΡΠΊΠΎΠ²ΠΈΠΌ ΡΠ°ΡΡΠΈΠ½Π°ΠΌ Π±Π°ΡΠ°Π»ΡΠΎΠ°ΠΌ ΡΠ΅ ΡΠΎΠ±ΠΈΠ»ΠΈ Π²ΡΠΎΡΠ°ΡΡΠΎΠ±ΡΠΏΡ ΡΠ°ΡΠΎΠΌ Ρ ΠΊΠΎΠΆΠ½ΠΎΠ³ΠΎ ΡΠΎΠΊΡ ΡΠΊΡΠΎ Π³ΠΎΠ²ΠΎΡΠΈΡΠΈ ΠΏΡΠΎ ΡΠΈΠ»ΠΈ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½Ρ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΠΎΠ±Π»Π°ΡΡΡ ΠΌΠΈ Π΄ΠΎΠΏΠΎΠΌΠΎΠΆΠ΅ ΠΌΠΎ ΡΡΡΠΌ ΡΠΈΠΌ Π±ΡΠ΄Π΅ ΠΏΠΎΡΡΡΠ±Π½ΠΎ ΠΌΠΎΠ²Π° ΠΏΡΠΎ Π³ΡΠΎΡΡ Π½Π΅ ΠΉΠ΄Π΅ ΠΌΠΈ Π½Π΅ ΠΏΡΠΎΡΡΠ½Π°Π½ΡΡΡΠΌΠΎ ΡΠ΅ΠΌΠΎΠ½ΡΡ ΡΠΈ ΡΠ½ΡΠΎΡ Π²ΡΠ»ΠΈΡΡ Π° Π½Π°ΡΠΈΠΌ Ρ
Π»ΠΎΠΏΡΡΠΌ Π΄ΠΎ ΠΏΠΎΠΌΠΎΠΆΠ΅ΠΌΠΎ Ρ ΡΠ΅ ΠΌΠ°Ρ ΡΠΎΠ±ΠΈΡΠΈ ΠΊΠΎΠ½Π° Π³ΡΠΎΠΌΠ°Π΄Π° Ρ ΠΊΠΎΠΆΠ΅Π½ Π»ΡΠ΄Π΅Ρ Π³ΠΌΠ°Π΄ΠΈ Π² Π½Π°ΡΡΠΉ ΠΊΡΠ°ΡΠ½Ρ
|
106 |
-
```
|
107 |
-
|
108 |
-
mer_lviv_interview.wav (with better News LM):
|
109 |
-
|
110 |
-
```
|
111 |
-
ΠΌΠΈ ΠΎΡΡΠΈΠΌΠ°Π»ΠΈ Π½ΠΎΠ²ΠΎΠ³ΠΎ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠ° ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΠΎΠ±Π»Π°ΡΡΡ Π° ΡΠ΅ Ρ ΡΠ°Ρ
ΠΎΠ²Π° Π»ΡΠ΄ΠΈΠ½Π° Ρ Π· ΡΡΡΠ½Π΅ ΠΌΡΡΡΠ° ΠΌΠΈ ΠΏΠ΅ΡΠ΅ΡΠΎΡΠΌΠ°ΡΡΠ²Π°Π»ΠΈ Π½Π°ΡΠ΅ ΡΠΏΡΠ°Π²Π»ΡΠ½Π½Ρ Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΈΡ
ΡΠΈΡΡΠ°ΡΡ ΡΡΠΎΠ³ΠΎΠ΄Π½Ρ Π²ΠΎΠ½ΠΎ ΠΌΠ°Ρ ΡΡΠ½ΠΊΡΡΡ Π½Π°Π΄Π·Π²ΠΈΡΠΉΠ½ΠΈΡ
ΡΠΈΡΡΠ°ΡΡ ΡΡΠ²ΡΠ»ΡΠ½ΠΎΠ³ΠΎ Π·Π°Ρ
ΠΈΡΡ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Ρ Π² Π½Π°Ρ Π΄ΠΎΡΡΠ°ΡΠ½ΡΠΎ ΡΠΊΡΡΠ½Π° ΡΠΏΡΠ²ΠΏΡΠ°ΡΡ ΡΠΊΡΡΠ½Π° ΠΊΠΎΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ ΡΠΎΠΌΡ ΡΠΎ Π·Π°Π΄Π°ΡΠ° ΠΌΡΡΡΠ° Π΄ΠΎΠΏοΏ½οΏ½ΠΌΠΎΠ³ΡΠΈ Π΄Π΅ΡΠΆΠ°Π²Π½ΠΎΠΌΡ ΡΠ°ΠΊΡΠΈΡΠ½ΠΎ ΠΊΠ΅ΡΡΠ½ΠΊΡ ΡΠΎΠ± Π²ΡΠ½ ΡΠΊΡΡΠ½ΠΎ Π·Π°ΠΏΡΠ°ΡΡΠ²Π°Π² ΡΡΠΎΡΠΎΠ²Π½ΠΎ ΡΠ΅ΠΌΠΎΠ½ΡΡ ΡΡΠΎΡΠΎΠ²Π½ΠΎ ΠΎΠ±Π»Π°Π΄Π½Π°Π½Π½Ρ Ρ ΡΡΠΎΡΠΎΠ²Π½ΠΎ ΡΠ΅ΠΊΡΡΡΠΈΠ½Π³Ρ Π»ΡΠ΄Π΅ΠΉ ΡΠΊΡ Π·Π°ΡΠ°Π· Π·Π³ΠΎΠ»ΠΎΡΡΠ²Π°Π»ΠΈΡΡ ΡΠΎΠ±ΠΈ Π±ΡΡΠΈ ΠΏΠΎ ΠΊΠΎΠ½ΡΡΠ°ΠΊΡΠ°Ρ
Π² ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΡΠΉ ΠΎΠ±ΠΎΡΠΎΠ½Ρ Π½Π° ΡΠ΅ΡΠΈΡΠΎΡΡΡ ΠΌΡΡΡΠ° Π»ΡΠ±ΠΎΠ²Π° Π½Π° ΡΡΠ²ΡΡΡΡΡΡΡ Π±Π°ΡΠ°Π»ΡΠΎΠ½ΠΈ Ρ Π½Π°ΡΠ°Π·Ρ Π²ΡΠ΅ ΠΉΠ΄Π΅ ΡΡΡΠΊΠΎ Π·Π° ΡΠΈΠΌ Π»Π°Π½ΠΎΠΌ ΡΠΊΡ Π²ΠΈΠ·Π½Π°ΡΠ΅Π½ΠΎΡ Π΄Π΅ΡΠΆΠ°Π²ΠΎΡ Π½Π°ΡΠΎΠΌΡΡΡΡ ΠΌΡΡΡΠΎ Π²Π·ΡΠ»ΠΎ Π½Π° ΡΠ΅Π±Π΅ Π΄ΠΎΠ΄Π°ΡΠΊΠΎΠ²Ρ ΡΡΠ½ΠΊΡΡΡ ΠΌΠΈ ΠΎΡΠ°Π»ΠΈ Π²Π΅Π»ΠΈΠΊΠΈΠΉ Π½Π°Π²ΡΠ°Π»ΡΠ½ΠΈΠΉ ΠΏΡΠΎΡΠ΅Ρ Π΄Π»Ρ ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΡΠ² ΠΊΠΎΠΌΡΠ½Π°Π»ΡΠ½ΠΈΡ
ΠΏΡΠ΄ΠΏΡΠΈΡΠΌΡΡΠΎΡΠΎΠ±Π»ΠΈΠ²ΠΎ ΠΌΠΎΠ² Π΄Π»Ρ ΠΏΡΠΎ ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½Ρ ΠΏΡΠ΄ΠΏΡΠΈΡΠΌΡΡΠ²Π° ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΠΈ Π²ΠΈΠΊΠΎΠ½Π°Π²ΡΠΈΠΉ Ρ
ΠΎΡΠ°Π½Ρ Π²Π»Π°Π΄ΠΈ Π±Π°Π½Π°Π»ΡΠ½ΠΎ Π²ΡΠ΄Π½ΠΎΠ²ΠΈΡΠΈ Π½Π°Π²ΠΈΠΊΠΈ Π²ΠΎΠ»ΠΎΠ΄ΡΠ½Π½Ρ Π·Π±ΡΠΎΡΡ Π² Π½Π°ΡΠΈΡ
ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΡΠ² ΠΏΠ»ΡΡ Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ ΡΠΈΠ½Π½ΠΈΠΊ ΡΠ΅ ΡΠΌΡΠ½Π½Ρ Π½Π°Π΄Π°Π²Π°ΡΠΈ ΠΏΠ΅ΡΡΡ Π΄ΠΎΠΌΠ΅Π΄ΠΈΡΠ½Ρ Π΄ΠΎΠΏΠΎΠΌΠΎΠ³Ρ Ρ Π·Π° ΡΠ°ΠΊΠ° Π²Π΅Π»ΠΈΠΊΠ° Ρ
Π²ΠΈΠ»Ρ ΡΠΊΠ° ΠΏΡΡΠ»Ρ Π·Π°ΠΊΡΠ½ΡΠ΅Π½Π½Ρ ΠΊΠ°ΡΠ°Π½ΡΠΈΠ²Π½ΠΎ Π±ΡΠ΄Π΅ Π²ΠΊΠ»ΡΡΠ°ΡΠΈ Ρ ΡΡΠ°ΡΡΠΎΠΊΠ»Π°ΡΠ½ΠΈΠΊΡΠ² ΠΌΠΈ Π½Π°ΠΏΡΠΈΠ²Π΅Π»ΠΈΠΊΠΈΠΉ ΠΆΠ°Π»Ρ Π±Π°Π³Π°ΡΠΎ Π² ΠΎΡΡΠ°Π½Π½ΡΡ
ΡΠΎΠΊΡΠ² ΡΠ°ΠΊΠ΅ Π²ΡΠ°ΠΆΠ΅Π½Π½Ρ ΡΠΎ ΠΏΠ΅ΡΠ΅Π±ΡΠ²Π°Π»ΠΈ Π»ΡΡΠΎΡΠΈΠ½ΠΎΠΌΡΡΠ½Ρ ΡΠΎΠ·ΡΠΌΡΡΡΠΈ ΡΠΎ ΡΠΎΡΡΡ ΠΏΠΎΡΡΡ Ρ Π·Π° Π²ΠΈΠ±ΡΠ΄Π΅ ΡΠ΅Π½ΡΡ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π°Π³ΡΠ΅ΡΡΡ Π½Π°ΠΌ ΡΡΠ΅Π±Π° Π²ΠΈΡΠΊΠΎΠ»ΡΠ²Π°ΡΠΈ ΡΠ΅Π±Π΅ Ρ Π±ΡΡΠΈ Π³ΠΎΡΠΎΠ²ΠΈΠΌΠΈ Π΄ΠΎ ΡΠΎΠ³ΠΎ ΡΠΎΠ± Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ ΡΡΠ΄Π½Ρ Π΄Π΅ΡΠΆΠ°Π²Ρ ΡΠΊΡΠΎ ΠΊΠΎΠΆΠ΅Π½ ΡΠΊΡΠ°ΡΠ½Π΅ΡΡ Π±ΡΠ΄Π΅ Π΄ΠΎΠ±ΡΠ΅ Π²ΠΎΠ»ΠΎΠ΄ΡΡΠΈ Π½Π°Π²ΠΈΠΊΠ°ΠΌΠΈ ΡΡΡΡΠ»Π΅ΡΡΠΊΠΎΡ Π·Π±ΡΠΎΡ Π±ΡΠ΄Π΅ Π²ΠΌΡΡΠΈ Π½Π°Π΄Π°Π²Π°ΡΠΈ ΠΌ Π΄ΠΈΡΠ½Ρ Π΄ΠΎΠΏΠΎΠΌΠΎΠ³Ρ Π² ΠΈΠΌΠ΅Π½ΠΈ ΡΠ΅ ΠΏΠΎΡΡΠΆΠ½Π° ΡΠΈΠ»Π° ΡΠΊΡ Π½Π΅ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π·Π΄ΠΎΠ»Π°ΡΠΈ ΡΠΎΠΌΡ Π² ΠΊΠΎΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ Π· Π΄Π΅ΡΠΆΠ°Π²Ρ Ρ ΡΡΡ
Π°ΡΠΌΠΎΡΡ Π²ΠΏΠ΅ΡΠ΅Π΄ ΡΠ΅ ΡΠ°ΠΊΠ΅ ΠΏΠΈΡΠ°Π½Π½Ρ ΠΎΡ Ρ Π½Π°Ρ Π±Π°Π³Π°ΡΠΎ ΠΊΠ°ΠΆΡΡΡ ΡΠΎ ΠΏΠΎΡΡΡ ΡΠ· Π±ΡΠΈΠ³Π°Π΄Π°ΠΌΠΈ ΡΠΈΠ» ΡΠ΅ΡΡΡΠ°Π»ΡΠΎΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΠΌΠΎΠΆΡΡΡ ΡΡΠ²ΡΡΠ²Π°ΡΠΈΡΡ Ρ ΡΠΎΡΠΌΡΠ²Π°ΡΠΈΡΡ Π΄ΠΎΠ±ΡΠΎΠ²ΠΎΠ»ΡΡΡ ΠΎΠ±ΡΠ΄Π½Π°Π½Π½Ρ ΡΠΈ ΡΠ²ΠΎΠ»Π²ΠΎΠ²Ρ ΡΠ΅ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ΅Π½ΠΎ ΡΠΊ Π²Π°ΡΠ΅ ΡΡΠ°Π²Π»Π΅Π½Π½Ρ Π΄ΠΎ ΡΠ°ΠΊΠΎΠ³ΠΎ Π²ΡΡΠ΅ ΠΌΠΈ ΠΏΡΠ°ΡΡΡΠΌΠΎ Π² ΡΡΡΠΊΠΈΠΉΠΊΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ Π· Π΄Π΅ΡΠΆΠ°Π²Π½ΠΈΠΌΠΈ ΡΠ½ΡΡΠΈΡΡΡΡΡΠΌΠΈ Ρ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½ΡΡΠ½ΡΠΉ Π΄Π΅Π½Ρ Ρ Π½Π΅ Π±Π°ΡΡ ΠΏΠΎΡΡΠ΅Π±ΠΈ Π² ΡΠ½ΡΠΈΡ
ΡΠΎΡΠΌΡΠ²Π°Π½Π½ΡΡ
ΡΠΎΠΌΡ ΡΠΎ Π² ΠΏΡΠΈΠ½ΡΠΈΠΏΡ ΠΌΠ°Ρ Π±ΡΡΠΈ ΠΆΠΎΡΡΡΠΊΠ° Π²Π΅ΡΡΠΈΠΊΠ°Π»Ρ ΠΌΠ°Ρ Π±ΡΡΠΈ ΠΎΠ΄ΠΈΠ½ ΡΠ΅Π½ΡΡ ΡΠΏΡΠ°Π²Π»ΡΠ½Π½Ρ ΡΠ»Π° ΠΎΠ±ΠΎ ΡΠΎ Ρ Π½ΠΎΠ²ΠΈΠΉ ΠΊΠΎΠΌΠ°Π½Π΄ΡΠ²Π°Ρ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π² ΡΠ°ΠΌΠΊΠ°Ρ
Π΄Π΅ΡΠΆΠ°Π²ΠΈ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊ ΡΠ°Π±Π°Π½ΠΎΠ²ΠΈ ΠΏΡΠΈΠ½ΡΠΈΠΏΡ ΠΏΠ°Π½Ρ ΡΠΎ ΡΠΊΡ Π² Π½Π°Ρ ΡΡΡΠ»ΠΈ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½Ρ ΠΎΡΠΎΠ½ ΠΎΠ±Π»Π°ΡΡΡ Π½Π° ΡΠ΅ Π΄ΡΠΆΠ΅ ΡΠ°Ρ
ΠΎΠ²Π° ΠΊΠΎΠΌΠΏΠ΅ΡΠ΅Π½ΡΠ½Π° Π»ΡΠ΄ΠΈΠ½Π° Π· Π½ΠΈΠΌ ΠΏΡΠΎΡΡΠΎ ΠΏΡΠΈΡΠΌΠ½ΠΎ ΡΠΏΡΠ²ΠΏΡΠ°ΡΡΠ²Π°ΡΠΈ Π½Π΅ ΡΠΎΠ·ΡΠΌΡΡ Π²ΡΠ΅ ΡΠΏΡΠ²ΡΠ»ΠΎΠ²Π° Π·Π°ΠΊΠΎΠ½ΠΎΠΌ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ΅Π½Π° ΠΌΠΎΠΆΠ»ΠΈΠ²ΡΡΡΡ ΠΏΠ΅ΡΠ΅ΠΊΠΈΠ΄Π°Π½Π½Ρ Π² ΡΠ°Π·Ρ ΡΠ°ΠΊΠΎΡ Π½Π°Π³Π°Π»ΡΠ½ΠΎΡ ΠΏΠΎΡΡΠ΅Π±ΠΈ ΠΏΠ΅ΡΠ΅ΠΊΠΈΠ΄Π°Π½Π½Ρ Π±ΡΠΈΠ³Π°Π΄ ΡΠΈΠ» ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π² Π·ΠΎΠ½Ρ Π±ΠΎΠΉΠΎΠ²ΠΈΡ
Π΄ΡΠΉ Π²ΡΠ½ΡΡΠΎΠ±Π»Π°ΡΡΡ ΡΠΊΠ΅ Π²Π°ΡΠ΅ ΡΡΠ°Π²Π»Π΅Π½Π½ Π΄ΠΎ ΡΡΠΎΠ³ΠΎ Ρ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ Π²Π°ΠΌ Π²ΡΠ΄ΠΎΠΌΠΎ ΡΠΊΠ΅ ΠΌΠΎΠΆΠ»ΠΈΠ²ΠΎ ΡΡΠ°Π²Π»Π΅Π½Π½Ρ ΡΡΡ ΡΠ°ΠΌΠΈΡ
Π±ΡΠΉΡΡΠ² ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎΡ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π²ΡΠ΄ ΡΠ΅ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½ΡΡΠ½ΡΠΉ Π΄Π΅Π½Ρ ΠΌΠΈ ΡΠΎΠ·ΡΠΌΡΡΠΌΠΎ ΡΠΎ Π²ΡΠ½ΡΠΎΡΠΌΠ°ΡΡΠΉΠ½ΠΎΠΌΡ ΠΏΠ»Π°Π½Ρ ΡΡΠ΅Π±Π° Π·ΡΠΎΠ±ΠΈΡΠΈ ΡΠΈΡΠ°Π½ΡΡΠ½ΠΈΠΉ ΠΎΠ±ΡΠΌ ΠΏΡΠ°ΡΡ Π»ΡΠ΄ΡΠΌ ΡΡΠ΅Π±Π° ΠΏΠΎΡΡΠ½ΠΈΡΠΈ ΡΠΎ ΡΠΈ Π½Π΅ ΠΌΠΎΠΆΠ΅Ρ ΡΡ
ΠΎΠ²Π°ΡΠΈΡΡ Ρ Π²Π»Π°ΡΠ½ΠΎΠΌΡ ΠΏΠΎΠΌΠ΅ΡΠΊΠ°Π½Π½Ρ ΠΊΠΎΠ»ΠΈ ΠΏΡΠΈΠΉΠ΄Π΅ Π²ΠΎΡΠΎΠ³ Π²ΠΎΡΠΎΠ³Π° ΡΡΠ΅Π±Π° Π½Π΅ ΠΏΡΡΡΠΈΡΠΈ Π½Π° ΡΠ΅ΡΠΈΡΠΎΡΡΡ Π½Π°ΡΠΎΡ ΠΊΡΠ°ΡΠ½ΠΈ Ρ Ρ Π΄ΡΠΌΠ°Ρ ΡΠΎ Π²ΡΡ Π³ΡΠΎΠΌΠ°Π΄ΡΠ½ΠΈ ΠΏΠΎΠ²ΠΈΠ½Π½Ρ ΠΌΠ°ΡΠΈ Π³ΠΎΡΠΎΠ²Π½Ρ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ ΡΠ²ΠΎΡ ΡΡΠ΄Π½Ρ ΠΊΡΡΠ½Ρ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½ΡΡΠ½ΡΠΉ Π΄Π΅Π½Ρ ΠΉΠ΄Π΅ Π½ΠΎΡΠΌΠ°Π»ΡΠ½ΠΈΠΉ ΠΏΡΠΎΡΠ΅Ρ Π»ΡΠ΄ΠΈ Π·Π°ΠΏΠΈΡΡΡΡΡΡΡ Π΄ΡΠΆΠ΅ Π±Π°Π³Π°ΡΠΎ Π²ΡΠ΄ΠΎΠΌΠ΅ Π»ΡΠ΄Π΅ΠΉ Π·Π°ΠΏΠΈΡΡΡΡΡΡΡ Π² ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½Ρ ΠΎΠ±ΠΎΡΠΎΠ½Ρ Ρ Π²Π²Π°ΠΆΠ°Ρ ΡΠ΅ ΠΏΡΠ°Π²ΠΈΠ»ΡΠ½ΠΎ ΡΠΎΠΌΡ ΡΠΎ ΡΡΠ΅Π±Π° ΠΌΠ°ΡΠΈ ΡΠ°ΠΊΡ Π²ΠΈΡΠΊΡΠ» Π° ΡΡΠΎΡΠΎΠ²Π½ΠΎ Π²ΡΡΡ
ΡΠ½ΡΠΈΡ
ΡΠ΅ΡΠ΅ΠΉ ΡΠ΅ ΡΠ°Π· Π½Π°Π³ΠΎΠ»ΠΎΡΡΡ ΠΊΠΎΠ»ΠΈ ΠΌΠΎΠ²Π° ΠΉΠ΄Π΅ ΠΏΡΠΎ Π±Π΅Π·ΠΏΠ΅ΠΊΡ ΠΏΡΠΎ ΠΎΠ±ΠΎΡΠΎΠ½Ρ Π΄Π΅ΡΠΆΠ°Π²ΠΈ ΠΌΠ°Ρ Π±ΡΡΠΈ ΡΡΠ±ΠΎΡΠ΄ΠΈΠ½Π°ΡΡΡ Ρ Ρ ΡΡΡΠΊΠ° Π΄Π΅ΡΠΆΠ°Π²Π½Π° Π²Π΅ΡΡΠΈΠΊΠ°Π»Ρ ΡΡΠ΅Π±Π° Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Π±ΡΠ΄Π΅ΠΌΠΎ ΡΡ ΡΠ°Π·ΠΎΠΌ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Π½Π΅ΠΌΠ° Π½Π° ΡΠΎ ΡΠ°Π΄ΠΈ Π° ΡΠ΅ ΠΎΠ΄Π½Π΅ ΠΏΠΈΡΠ°Π½Π½Ρ Π±Π°Π³Π°ΡΠΎ Ρ
ΡΠΎ Π²ΠΏΠ΅Π²Π½Π΅Π½οΏ½οΏ½ΠΉ ΡΠΎ Π² Π·Π°ΠΊΠΎΠ½Ρ Π½Π΅Π΄ΠΎΡΡΠ°ΡΠ½ΡΠΎ Π²Π»Π°Π΄ΠΈ ΠΏΡΠΎΠΏΠΈΡΠ°Π½ΠΎ Π΄Π»Ρ ΠΌΡΡΡΠ΅Π²ΠΈΡ
Π΄ ΠΌΡΠ½ΡΡΡΡΠ°ΡΡΡ Π²Π»Π΄ΠΈ ΠΌΠ°ΡΡΡΡΡ Π½Π° ΡΠ²Π°Π·Ρ ΠΏΠΎΠ²Π½ΠΎΠ²Π°ΠΆΠ΅Π½Ρ Ρ ΠΌΠΎΠΆΠ»Π²ΠΎΡΡΠΏΡΠΈ ΡΠΎΡΠΌΡΠ²Π°Π½Π½Ρ ΡΠΎΠ·Π³ΠΎΡΡΠ°Π½Π½Ρ ΡΠΈΠ» ΡΠΎΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π²ΠΈ Π½Π° ΠΏΡΠ°ΠΊΡΠΈΡΡ Π·ΡΡΠΊΠ½ΡΠ»ΠΈΡΡ Π²ΠΆΠ΅ ΡΠΈΠΌ Ρ ΡΠΊ Π²ΠΈ ΠΎΡΡΠ½ΡΡΡΡΡΡΡΠ΅Π½Ρ ΠΏΠΎΠΏΠ΅ΡΡΠ΅ Π΄ΠΎΠ±ΡΠ΅ ΡΠΎ Π·Π°ΠΊΠΎΠ½ Ρ Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎ ΡΠΎ ΠΆΠΎΠ΄ΠΎΠ΄Π΅Π½Π·Π·Π°ΠΎΠ½ Π½Π΅ Ρ ΡΠ° ΠΎΠ±ΡΠ°Π»ΡΠ½ΠΈΠΉ ΡΠΈ ΡΠ΄Π΅Π°Π»ΡΠ½ΠΈΠΉ Π²ΡΠ½ ΠΌΠ°Π² Π±ΠΈ ΠΏΡΠΎΡΠ΅ΡΡ ΡΠ΅Π°Π»ΡΠ½ΠΎ ΠΆΠΈΡΡΡ Π·Π°Π·Π½Π°ΡΠΈ ΠΏΠ΅Π²Π½ΠΈΡ
ΠΊΠΎΡΠΈΠ³ΡΠ²Π°Π½Ρ Π½Ρ Π½Π°ΠΏΡΠΈΠΊΠ»Π°Π΄ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠΎΠΌ ΡΠ΅ΡΠΈΡΡΠ°Π»ΡΠ½ΠΎΠΉ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ Π² ΡΠ°ΡΡ Π½Π° Π·Π²ΠΈΡΠ°ΠΉΠ½ΠΎΡ ΡΠΈΡΡΠ°ΡΡΡ Ρ Π³ΠΎΠ»ΠΎΠ²Π° ΠΎΠ±Π»Π°ΡΠ½ΠΎΡ Π°Π΄ΠΌΡΠ½ΡΡΡΡΠ°ΡΡΡ ΡΠ°ΠΊ Π° ΠΏΠΎ ΠΌΡΡΡΡ Π»ΡΠ±ΠΎΠ²ΠΎΠ½Π°ΠΏΡΠΈΠΊΠ»Π°Π΄ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠΎΠΌ Ρ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊ ΡΠ°ΠΉΠΎΠ½Π½ΠΎΡ Π°Π΄ΠΌΡΠ½ΡΡΡΡΠ°ΡΡΡ ΡΠ΅ Ρ Π½Π΅Π²Π΅Π»ΠΈΠΊΠΈΠΉ ΠΎΡΡΡ Π½Π΅Ρ
Π΅Π΄Π΅ΠΊΡΠ»ΡΠΊΠ° ΠΏΡΠ°ΡΡΠ²Π½ΠΈΠΊΡΠ² Ρ Π²ΠΎΠ½ΠΈ ΠΌΠ°Π»ΠΈ Π±ΠΈ ΡΡΠΎΡΡΠΈ Π½Π°Π΄ Π³ΡΠΎΠΌΠ°Π΄ΠΎΡ ΠΌΡΡΡΠ° Π»ΡΠ²ΠΎΠ²Π° Ρ Π½Π΅ Π΄ΡΠΌΠ°Ρ ΡΠΎ ΡΠ΅ Π±ΡΠ΄Π΅ ΠΌΠ°ΡΠΈ Π΄ΠΎΡΡΠ°ΡΠ½Ρ ΡΠΊΡΡΠ½ΠΈΠΉ Π΅ΡΠ΅ΠΊΡ ΡΠΎΠΌΡ ΡΠΎ ΠΏΡΠΈΠ½ΡΠΈΠΏ Ρ ΠΊΠ΅ΡΡΠ²Π½ΠΈΠΊΠΎΠ±Π»Π°ΡΡΡ ΠΊΠ΅ΡΡΠ²Π½ΠΈ ΠΌΡΡΠ° Π°Π»Π΅ ΡΡ ΡΠ΅ΡΡ Π½Π°ΠΏΠ΅Π²Π½ΠΎ ΠΊΠΎΠ»ΠΈ ΠΏΠΈΡΠ°Π»ΠΈ Π·Π°ΠΊΠΎΠ½ Π½Π΅ Π΄ΠΎ ΠΊΡΠ½ΡΡ ΡΠ°Ρ
ΠΎΠ²ΡΠ²Π°Π»ΠΈ Π°Π»Π΅ ΠΌΠΈ ΠΌΠ°ΡΠΌΠΎ ΡΠ°Ρ ΡΠΎΠ± ΡΠ΅ ΠΏΠΎΠΏΡΠ°Π²ΠΈΡΠΈ Π°Π»Π΅ Π½Π° ΡΡΠΎΠ³ΠΎΠ΄Π½Ρ ΠΌΠΈ ΡΡΡΠΊΠΎ Π²ΠΈΠΊΠΎΠ½ΠΈΠΎΠ·Π°ΠΊΠ½ΠΈ ΡΠΊΠΈΠΉ Ρ ΡΠ½ΡΡ
Π²Π°ΡΡΠ½ΡΡΠ² Π½Π΅ ΠΌΠΎΠΆΠ΅ Π±ΡΠ΄Π΅ ΠΏΠ°Π½Π΅ Π°Π½Π΄ΡΡΡ Π±ΡΠΊΠ²Π°Π»ΡΠ½ΠΎ ΠΎΡΡΠ°Π½Π½Ρ Π·Π°ΠΏΠΈΡΠ½Π½Ρ Π²Π°ΡΠ° Π±ΡΠΈΠ³Π°Π΄Π° Π»ΡΠ²ΡΠ²ΡΡΠΊΠ° Π½Ρ Π΄Π²Π° Π±Π°ΡΠ°Π»ΡΠΉΠΎΠ½ΠΈ ΡΠΊ Π²ΠΈΠΊΡΠΈΡΠ° ΡΠΊ Π²ΠΎΠ½ΠΈ Π·Π°ΡΠ°Π· ΠΎΠ·Π±ΡΠΎΡΠ½Ρ ΠΏΠΎΠ²Π½ΡΡΡΡ ΡΠΈ Π½ΡΡ ΡΠ΅ ΡΡΠ»ΡΠΊΠΈ ΡΡΡΡΠ»Π΅ΡΡΠΊΠ΅ Π·Π±ΡΡ ΡΠΈ ΡΠ΅ ΠΉ Π±ΡΠ»ΡΡ Π²Π°ΠΆΠΊΠ΅ ΠΎΠ·Π±ΡΠΎΡΠ½Π½Ρ Ρ ΡΠΊ Π² ΡΡΠ°Π²ΠΈΡΠΈΡΡ Π΄ΠΎ ΡΠΎΠ³ΠΎ ΡΠΎΠ± ΠΌΡΡΡΠΊΠ΄ΡΡΡΡΠ°ΡΡΡ ΠΌΠ°Π»Π° ΠΌΠΎΠΆΠ»ΠΈΠ²ΡΡΡΡΠΊΠ°ΠΆΡΠΌΠΎ ΡΠΎΡΡ ΠΊΡΠΏΡΠ²Π°ΡΠΈ Π΄Π»Ρ Π·Π°Π±Π·ΠΏΠ΅ΡΠ΅Π½Π½Ρ ΡΠ°ΠΌΠ΅ ΡΠΈ ΡΡΡΡΠ°Π»ΡΠΎΠΎΡΠΎΠ½ΠΈ Π½Ρ Π½Π°ΠΏΡΠΈΠΊΠ»Π°Π΄ ΡΡ ΠΆ ΡΠ°ΠΌΡ Π±Π΅ΡΠΏΡΠ»ΠΎΡΠ½ΠΈΠΊΠΈ Π΄Π»Ρ ΡΠΎΠ·Π²ΡΠ΄ΠΊΠΈ Π°Π±ΠΎ Π·Π°ΡΠΎΠ±ΠΈ Π·Π²ΡΠ·ΠΊΡ Π° Π°Π±ΠΎ ΠΎΡΡ ΡΠΎΡΡ ΡΠ°ΠΊΠ΅ Π½Ρ ΡΠΎΠ± Π²ΠΈ ΡΠΎΠ·ΡΠΌΡΠ»ΠΈ ΠΌΠΈ Π΄ΠΎΡΡΠ°ΡΠ½ΡΠΎ Π±Π°Π³Π°ΡΠΎ ΠΏΠΎΠΌΠ°Π³Π°ΡΠΌΠΎ Π²ΡΡΠΌ Π½Π°ΡΠΈΠΌ Π²ΡΠΉΡΡΠΊΠΎΠ²ΠΈΠΌ ΡΠ°ΡΡΠΈΠ½Π°ΠΌ Π±Π°ΡΠ°Π»ΡΠΎΠ½Π°ΠΌ ΡΠ΅ ΡΠΎΠ±ΠΈΠ»ΠΈ ΡΠΈΡΠ΅ΡΠ°ΡΡΠΎΠ±Ρ ΠΏΠΈΡΠ°ΡΠΈΠΌΡ ΠΊΠΎΠΆΠ½ΠΎΠ³ΠΎ ΡΠΎΠΊΡ ΡΠΊΡΠΎ Π³ΠΎΠ²ΠΎΡΠΈΡΠΈ ΠΏΡΠΎ ΡΠΈΠ»ΠΈ ΡΠ΅ΡΠΈΡΠΎΡΡΠ°Π»ΡΠ½ΠΎ ΠΎΠ±ΠΎΡΠΎΠ½ΠΈ ΠΎΠ±Π»Π°ΡΡΡ ΠΌΠΈ Π΄ΠΎΠΏΠΎΠΌΠΎΠΆΠ΅ΠΌΠΎ ΡΡΡΠΌ ΡΠΈΠΌ Π±ΡΠ΄Π΅ ΠΏΠΎΡΡΡΠ±Π½ΠΎ ΠΌΠΎΠ²Π° ΠΏΡΠΎ Π³ΡΠΎΡΡ Π½Π΅ ΠΉΠ΄Π΅ ΠΌΠΈ Π½Π΅ ΠΏΡΠΎΡΡΠ½Π°Π½ΡΡΡΠΌΠΎ ΡΠ΅ΠΌΠΎΠ½ΡΡΡΡΠΈ ΡΠ½ΡΠΎΡ Π²ΡΠ»ΠΈΡΡ Π° Π½Π°ΡΠΈΠΌ Ρ
Π»ΠΎΠΏΡΡΠΌ Π΄ΠΎΠΏΠΎΠΌΠΎΠΆΠ΅ΠΌΠΎ Ρ ΡΠ΅ ΠΌΠ°Ρ ΡΠΎΠ±ΠΈΡΠΈ ΠΊΠΎΠ½Π° Π³ΡΠΎΠΌΠ°Π΄Π° Ρ ΠΊΠΎΠΆΠ΅Π½ Π»ΡΠ΄Π΅Ρ Π³ΠΎΠΌΠ°Π΄ΠΈ Π² Π½Π°ΡΡΠΉ ΠΊΡΠ°ΡΠ½Ρ
|
112 |
-
```
|
113 |
-
|
114 |
-
### Inference of `mer_lviv_interview.wav` (time is 06:38)
|
115 |
-
|
116 |
-
#### CPU
|
117 |
-
|
118 |
-
- Memory peaks to 60GB
|
119 |
-
- Memory peaks to 65GB (on News LM)
|
120 |
-
|
121 |
-
Inference duration:
|
122 |
-
|
123 |
-
```
|
124 |
-
real 7m39.461s
|
125 |
-
user 59m19.065s
|
126 |
-
sys 24m1.254s
|
127 |
-
```
|
128 |
-
|
129 |
-
Inference duration (on News LM):
|
130 |
-
|
131 |
-
```
|
132 |
-
real 12m36.888s
|
133 |
-
user 63m19.396s
|
134 |
-
sys 24m24.823s
|
135 |
-
```
|
136 |
-
|
137 |
-
Duration tracked with loading the LM.
|
138 |
-
|
139 |
-
## Using timestamps
|
140 |
-
|
141 |
-
The `inference_timestamps.py` script can be used to do inference with timestamps for chars and words.
|
142 |
-
|
143 |
-
### `output_char_offsets=True`
|
144 |
-
|
145 |
-
```
|
146 |
-
Wav2Vec2CTCTokenizerOutput(text='ΠΏΠ°Π½Ρ ΡΠΏΠΎΠ»ΡΡΠ΅Π½Ρ ΡΡΠ°ΡΠΈ Π½Π°Π΄Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½ΠΈΠΉ ΠΏΠ°ΡΡΠ½Π΅Ρ ΠΎΠ΄Π½Π°ΠΊ Ρ ΡΡΠ·Π½ΠΈΡΡ ΡΡΠ°ΡΠΈ ΠΌΠ°ΡΡΡ ΡΠΏΠ΅ΡΡΠ°Π»ΡΠ½ΠΈ Π·Π°ΠΊΠΎΠ½ ΡΠΊΠΈΠΉ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ°Ρ ΡΠΊΡΠΎ ΠΊΠΈΡΠ°ΠΉ Π½Π°ΠΏΠ°Π΄Π΅ Π½Π° ΡΠ°ΠΉΠ²Π°Π½Ρ Π°ΠΌΠ΅ΡΠΈΠΊΠ°Π½ΡΡΠΊΠΈΠΉ Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ ΠΌΠ°ΡΡΡ ΠΉΠΎΠ³ΠΎ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Π΅Π²ΡΠΉΠ²Π³Π΅ΡΠ΅', char_offsets=[{'char': 'ΠΏ', 'start_offset': 0, 'end_offset': 1}, {'char': 'Π°', 'start_offset': 1, 'end_offset': 2}, {'char': 'Π½', 'start_offset': 9, 'end_offset': 10}, {'char': 'Ρ', 'start_offset': 11, 'end_offset': 12}, {'char': ' ', 'start_offset': 14, 'end_offset': 15}, {'char': 'Ρ', 'start_offset': 16, 'end_offset': 17}, {'char': 'ΠΏ', 'start_offset': 19, 'end_offset': 20}, {'char': 'ΠΎ', 'start_offset': 21, 'end_offset': 22}, {'char': 'Π»', 'start_offset': 23, 'end_offset': 24}, {'char': 'Ρ', 'start_offset': 25, 'end_offset': 26}, {'char': 'Ρ', 'start_offset': 30, 'end_offset': 31}, {'char': 'Π΅', 'start_offset': 32, 'end_offset': 33}, {'char': 'Π½', 'start_offset': 37, 'end_offset': 38}, {'char': 'Ρ', 'start_offset': 38, 'end_offset': 39}, {'char': ' ', 'start_offset': 40, 'end_offset': 42}, {'char': 'Ρ', 'start_offset': 43, 'end_offset': 44}, {'char': 'Ρ', 'start_offset': 46, 'end_offset': 47}, {'char': 'Π°', 'start_offset': 48, 'end_offset': 49}, {'char': 'Ρ', 'start_offset': 57, 'end_offset': 58}, {'char': 'ΠΈ', 'start_offset': 58, 'end_offset': 59}, {'char': ' ', 'start_offset': 76, 'end_offset': 79}, {'char': 'Π½', 'start_offset': 85, 'end_offset': 86}, {'char': 'Π°', 'start_offset': 87, 'end_offset': 88}, {'char': 'Π΄', 'start_offset': 93, 'end_offset': 94}, {'char': 'Π²', 'start_offset': 97, 'end_offset': 98}, {'char': 'Π°', 'start_offset': 99, 'end_offset': 100}, {'char': 'ΠΆ', 'start_offset': 105, 'end_offset': 106}, {'char': 'Π»', 'start_offset': 113, 'end_offset': 114}, {'char': 'ΠΈ', 'start_offset': 114, 'end_offset': 115}, {'char': 'Π²', 'start_offset': 121, 'end_offset': 122}, {'char': 'ΠΈ', 'start_offset': 123, 'end_offset': 124}, {'char': 'ΠΉ', 'start_offset': 125, 'end_offset': 126}, {'char': ' ', 'start_offset': 127, 'end_offset': 129}, {'char': 'Ρ', 'start_offset': 130, 'end_offset': 131}, {'char': 'Ρ', 'start_offset': 134, 'end_offset': 136}, {'char': 'Ρ', 'start_offset': 138, 'end_offset': 139}, {'char': 'Π°', 'start_offset': 139, 'end_offset': 140}, {'char': 'Ρ', 'start_offset': 145, 'end_offset': 146}, {'char': 'Π΅', 'start_offset': 146, 'end_offset': 147}, {'char': 'Π³', 'start_offset': 152, 'end_offset': 153}, {'char': 'Ρ', 'start_offset': 153, 'end_offset': 154}, {'char': 'Ρ', 'start_offset': 160, 'end_offset': 161}, {'char': 'Π½', 'start_offset': 167, 'end_offset': 168}, {'char': 'ΠΈ', 'start_offset': 168, 'end_offset': 169}, {'char': 'ΠΉ', 'start_offset': 170, 'end_offset': 171}, {'char': ' ', 'start_offset': 171, 'end_offset': 173}, {'char': 'ΠΏ', 'start_offset': 174, 'end_offset': 175}, {'char': 'Π°', 'start_offset': 176, 'end_offset': 177}, {'char': 'Ρ', 'start_offset': 179, 'end_offset': 180}, {'char': 'Ρ', 'start_offset': 183, 'end_offset': 184}, {'char': 'Π½', 'start_offset': 188, 'end_offset': 189}, {'char': 'Π΅', 'start_offset': 189, 'end_offset': 190}, {'char': 'Ρ', 'start_offset': 193, 'end_offset': 194}, {'char': ' ', 'start_offset': 201, 'end_offset': 203}, {'char': 'ΠΎ', 'start_offset': 204, 'end_offset': 205}, {'char': 'Π΄', 'start_offset': 208, 'end_offset': 209}, {'char': 'Π½', 'start_offset': 214, 'end_offset': 216}, {'char': 'Π°', 'start_offset': 216, 'end_offset': 217}, {'char': 'ΠΊ', 'start_offset': 224, 'end_offset': 225}, {'char': ' ', 'start_offset': 227, 'end_offset': 229}, {'char': 'Ρ', 'start_offset': 233, 'end_offset': 234}, {'char': ' ', 'start_offset': 237, 'end_offset': 239}, {'char': 'Ρ', 'start_offset': 240, 'end_offset': 241}, {'char': 'Ρ', 'start_offset': 241, 'end_offset': 242}, {'char': 'Π·', 'start_offset': 247, 'end_offset': 248}, {'char': 'Π½', 'start_offset': 253, 'end_offset': 254}, {'char': 'ΠΈ', 'start_offset': 254, 'end_offset': 255}, {'char': 'Ρ', 'start_offset': 261, 'end_offset': 262}, {'char': 'Ρ', 'start_offset': 262, 'end_offset': 263}, {'char': ' ', 'start_offset': 281, 'end_offset': 283}, {'char': 'Ρ', 'start_offset': 283, 'end_offset': 284}, {'char': 'Ρ', 'start_offset': 286, 'end_offset': 287}, {'char': 'Π°', 'start_offset': 288, 'end_offset': 289}, {'char': 'Ρ', 'start_offset': 294, 'end_offset': 295}, {'char': 'ΠΈ', 'start_offset': 296, 'end_offset': 297}, {'char': ' ', 'start_offset': 297, 'end_offset': 299}, {'char': 'ΠΌ', 'start_offset': 300, 'end_offset': 301}, {'char': 'Π°', 'start_offset': 301, 'end_offset': 302}, {'char': 'Ρ', 'start_offset': 306, 'end_offset': 307}, {'char': 'Ρ', 'start_offset': 308, 'end_offset': 309}, {'char': 'Ρ', 'start_offset': 309, 'end_offset': 311}, {'char': ' ', 'start_offset': 311, 'end_offset': 313}, {'char': 'Ρ', 'start_offset': 313, 'end_offset': 314}, {'char': 'ΠΏ', 'start_offset': 316, 'end_offset': 317}, {'char': 'Π΅', 'start_offset': 318, 'end_offset': 319}, {'char': 'Ρ', 'start_offset': 324, 'end_offset': 325}, {'char': 'Ρ', 'start_offset': 325, 'end_offset': 326}, {'char': 'Π°', 'start_offset': 328, 'end_offset': 329}, {'char': 'Π»', 'start_offset': 333, 'end_offset': 334}, {'char': 'Ρ', 'start_offset': 334, 'end_offset': 336}, {'char': 'Π½', 'start_offset': 339, 'end_offset': 340}, {'char': 'ΠΈ', 'start_offset': 341, 'end_offset': 342}, {'char': ' ', 'start_offset': 345, 'end_offset': 348}, {'char': 'Π·', 'start_offset': 351, 'end_offset': 352}, {'char': 'Π°', 'start_offset': 354, 'end_offset': 355}, {'char': 'ΠΊ', 'start_offset': 361, 'end_offset': 362}, {'char': 'ΠΎ', 'start_offset': 365, 'end_offset': 366}, {'char': 'Π½', 'start_offset': 373, 'end_offset': 374}, {'char': ' ', 'start_offset': 382, 'end_offset': 384}, {'char': 'Ρ', 'start_offset': 386, 'end_offset': 387}, {'char': 'ΠΊ', 'start_offset': 390, 'end_offset': 391}, {'char': 'ΠΈ', 'start_offset': 392, 'end_offset': 393}, {'char': 'ΠΉ', 'start_offset': 394, 'end_offset': 395}, {'char': ' ', 'start_offset': 396, 'end_offset': 398}, {'char': 'ΠΏ', 'start_offset': 399, 'end_offset': 401}, {'char': 'Π΅', 'start_offset': 402, 'end_offset': 403}, {'char': 'Ρ', 'start_offset': 406, 'end_offset': 407}, {'char': 'Π΅', 'start_offset': 407, 'end_offset': 408}, {'char': 'Π΄', 'start_offset': 411, 'end_offset': 412}, {'char': 'Π±', 'start_offset': 415, 'end_offset': 416}, {'char': 'Π°', 'start_offset': 416, 'end_offset': 417}, {'char': 'Ρ', 'start_offset': 424, 'end_offset': 425}, {'char': 'Π°', 'start_offset': 428, 'end_offset': 429}, {'char': 'Ρ', 'start_offset': 437, 'end_offset': 438}, {'char': ' ', 'start_offset': 445, 'end_offset': 447}, {'char': 'Ρ', 'start_offset': 448, 'end_offset': 449}, {'char': 'ΠΊ', 'start_offset': 452, 'end_offset': 453}, {'char': 'Ρ', 'start_offset': 455, 'end_offset': 456}, {'char': 'ΠΎ', 'start_offset': 457, 'end_offset': 458}, {'char': ' ', 'start_offset': 460, 'end_offset': 463}, {'char': 'ΠΊ', 'start_offset': 463, 'end_offset': 464}, {'char': 'ΠΈ', 'start_offset': 465, 'end_offset': 466}, {'char': 'Ρ', 'start_offset': 470, 'end_offset': 471}, {'char': 'Π°', 'start_offset': 472, 'end_offset': 473}, {'char': 'ΠΉ', 'start_offset': 478, 'end_offset': 480}, {'char': ' ', 'start_offset': 484, 'end_offset': 486}, {'char': 'Π½', 'start_offset': 487, 'end_offset': 488}, {'char': 'Π°', 'start_offset': 488, 'end_offset': 489}, {'char': 'ΠΏ', 'start_offset': 493, 'end_offset': 494}, {'char': 'Π°', 'start_offset': 496, 'end_offset': 497}, {'char': 'Π΄', 'start_offset': 502, 'end_offset': 503}, {'char': 'Π΅', 'start_offset': 504, 'end_offset': 505}, {'char': ' ', 'start_offset': 509, 'end_offset': 511}, {'char': 'Π½', 'start_offset': 511, 'end_offset': 512}, {'char': 'Π°', 'start_offset': 513, 'end_offset': 514}, {'char': ' ', 'start_offset': 515, 'end_offset': 517}, {'char': 'Ρ', 'start_offset': 518, 'end_offset': 519}, {'char': 'Π°', 'start_offset': 519, 'end_offset': 520}, {'char': 'ΠΉ', 'start_offset': 524, 'end_offset': 525}, {'char': 'Π²', 'start_offset': 527, 'end_offset': 528}, {'char': 'Π°', 'start_offset': 529, 'end_offset': 530}, {'char': 'Π½', 'start_offset': 535, 'end_offset': 536}, {'char': 'Ρ', 'start_offset': 536, 'end_offset': 537}, {'char': ' ', 'start_offset': 552, 'end_offset': 555}, {'char': 'Π°', 'start_offset': 555, 'end_offset': 556}, {'char': 'ΠΌ', 'start_offset': 561, 'end_offset': 562}, {'char': 'Π΅', 'start_offset': 562, 'end_offset': 563}, {'char': 'Ρ', 'start_offset': 566, 'end_offset': 567}, {'char': 'ΠΈ', 'start_offset': 567, 'end_offset': 568}, {'char': 'ΠΊ', 'start_offset': 572, 'end_offset': 573}, {'char': 'Π°', 'start_offset': 574, 'end_offset': 575}, {'char': 'Π½', 'start_offset': 579, 'end_offset': 580}, {'char': 'Ρ', 'start_offset': 582, 'end_offset': 583}, {'char': 'Ρ', 'start_offset': 583, 'end_offset': 585}, {'char': 'ΠΊ', 'start_offset': 586, 'end_offset': 587}, {'char': 'ΠΈ', 'start_offset': 588, 'end_offset': 589}, {'char': 'ΠΉ', 'start_offset': 589, 'end_offset': 590}, {'char': ' ', 'start_offset': 591, 'end_offset': 593}, {'char': 'Π²', 'start_offset': 594, 'end_offset': 595}, {'char': 'Ρ', 'start_offset': 595, 'end_offset': 596}, {'char': 'ΠΉ', 'start_offset': 600, 'end_offset': 601}, {'char': 'Ρ', 'start_offset': 604, 'end_offset': 605}, {'char': 'Ρ', 'start_offset': 605, 'end_offset': 607}, {'char': 'ΠΊ', 'start_offset': 609, 'end_offset': 611}, {'char': 'ΠΎ', 'start_offset': 612, 'end_offset': 613}, {'char': 'Π²', 'start_offset': 620, 'end_offset': 621}, {'char': 'Ρ', 'start_offset': 622, 'end_offset': 623}, {'char': ' ', 'start_offset': 637, 'end_offset': 639}, {'char': 'ΠΌ', 'start_offset': 641, 'end_offset': 642}, {'char': 'Π°', 'start_offset': 643, 'end_offset': 644}, {'char': 'Ρ', 'start_offset': 651, 'end_offset': 652}, {'char': 'Ρ', 'start_offset': 654, 'end_offset': 655}, {'char': 'Ρ', 'start_offset': 655, 'end_offset': 656}, {'char': ' ', 'start_offset': 657, 'end_offset': 659}, {'char': 'ΠΉ', 'start_offset': 659, 'end_offset': 660}, {'char': 'ΠΎ', 'start_offset': 660, 'end_offset': 662}, {'char': 'Π³', 'start_offset': 664, 'end_offset': 665}, {'char': 'ΠΎ', 'start_offset': 666, 'end_offset': 667}, {'char': ' ', 'start_offset': 677, 'end_offset': 679}, {'char': 'Π·', 'start_offset': 681, 'end_offset': 682}, {'char': 'Π°', 'start_offset': 683, 'end_offset': 684}, {'char': 'Ρ
', 'start_offset': 686, 'end_offset': 687}, {'char': 'ΠΈ', 'start_offset': 689, 'end_offset': 690}, {'char': 'Ρ', 'start_offset': 696, 'end_offset': 697}, {'char': 'Π°', 'start_offset': 698, 'end_offset': 699}, {'char': 'Ρ', 'start_offset': 707, 'end_offset': 708}, {'char': 'ΠΈ', 'start_offset': 709, 'end_offset': 710}, {'char': ' ', 'start_offset': 733, 'end_offset': 734}, {'char': 'Π΅', 'start_offset': 740, 'end_offset': 741}, {'char': 'Π²', 'start_offset': 747, 'end_offset': 748}, {'char': 'Ρ', 'start_offset': 748, 'end_offset': 749}, {'char': 'ΠΉ', 'start_offset': 752, 'end_offset': 753}, {'char': 'Π²', 'start_offset': 754, 'end_offset': 755}, {'char': 'Π³', 'start_offset': 757, 'end_offset': 758}, {'char': 'Π΅', 'start_offset': 759, 'end_offset': 760}, {'char': 'Ρ', 'start_offset': 767, 'end_offset': 768}, {'char': 'Π΅', 'start_offset': 768, 'end_offset': 769}], word_offsets=None)
|
147 |
-
```
|
148 |
-
|
149 |
-
### `output_word_offsets=True`
|
150 |
-
|
151 |
-
```
|
152 |
-
Wav2Vec2CTCTokenizerOutput(text='ΠΏΠ°Π½Ρ ΡΠΏΠΎΠ»ΡΡΠ΅Π½Ρ ΡΡΠ°ΡΠΈ Π½Π°Π΄Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½ΠΈΠΉ ΠΏΠ°ΡΡΠ½Π΅Ρ ΠΎΠ΄Π½Π°ΠΊ Ρ ΡΡΠ·Π½ΠΈΡΡ ΡΡΠ°ΡΠΈ ΠΌΠ°ΡΡΡ ΡΠΏΠ΅ΡΡΠ°Π»ΡΠ½ΠΈ Π·Π°ΠΊΠΎΠ½ ΡΠΊΠΈΠΉ ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ°Ρ ΡΠΊΡΠΎ ΠΊΠΈΡΠ°ΠΉ Π½Π°ΠΏΠ°Π΄Π΅ Π½οΏ½οΏ½ ΡΠ°ΠΉΠ²Π°Π½Ρ Π°ΠΌΠ΅ΡΠΈΠΊΠ°Π½ΡΡΠΊΠΈΠΉ Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ ΠΌΠ°ΡΡΡ ΠΉΠΎΠ³ΠΎ Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ Π΅Π²ΡΠΉΠ²Π³Π΅ΡΠ΅', char_offsets=[{'char': 'ΠΏ', 'start_offset': 0, 'end_offset': 1}, {'char': 'Π°', 'start_offset': 1, 'end_offset': 2}, {'char': 'Π½', 'start_offset': 9, 'end_offset': 10}, {'char': 'Ρ', 'start_offset': 11, 'end_offset': 12}, {'char': ' ', 'start_offset': 14, 'end_offset': 15}, {'char': 'Ρ', 'start_offset': 16, 'end_offset': 17}, {'char': 'ΠΏ', 'start_offset': 19, 'end_offset': 20}, {'char': 'ΠΎ', 'start_offset': 21, 'end_offset': 22}, {'char': 'Π»', 'start_offset': 23, 'end_offset': 24}, {'char': 'Ρ', 'start_offset': 25, 'end_offset': 26}, {'char': 'Ρ', 'start_offset': 30, 'end_offset': 31}, {'char': 'Π΅', 'start_offset': 32, 'end_offset': 33}, {'char': 'Π½', 'start_offset': 37, 'end_offset': 38}, {'char': 'Ρ', 'start_offset': 38, 'end_offset': 39}, {'char': ' ', 'start_offset': 40, 'end_offset': 42}, {'char': 'Ρ', 'start_offset': 43, 'end_offset': 44}, {'char': 'Ρ', 'start_offset': 46, 'end_offset': 47}, {'char': 'Π°', 'start_offset': 48, 'end_offset': 49}, {'char': 'Ρ', 'start_offset': 57, 'end_offset': 58}, {'char': 'ΠΈ', 'start_offset': 58, 'end_offset': 59}, {'char': ' ', 'start_offset': 76, 'end_offset': 79}, {'char': 'Π½', 'start_offset': 85, 'end_offset': 86}, {'char': 'Π°', 'start_offset': 87, 'end_offset': 88}, {'char': 'Π΄', 'start_offset': 93, 'end_offset': 94}, {'char': 'Π²', 'start_offset': 97, 'end_offset': 98}, {'char': 'Π°', 'start_offset': 99, 'end_offset': 100}, {'char': 'ΠΆ', 'start_offset': 105, 'end_offset': 106}, {'char': 'Π»', 'start_offset': 113, 'end_offset': 114}, {'char': 'ΠΈ', 'start_offset': 114, 'end_offset': 115}, {'char': 'Π²', 'start_offset': 121, 'end_offset': 122}, {'char': 'ΠΈ', 'start_offset': 123, 'end_offset': 124}, {'char': 'ΠΉ', 'start_offset': 125, 'end_offset': 126}, {'char': ' ', 'start_offset': 127, 'end_offset': 129}, {'char': 'Ρ', 'start_offset': 130, 'end_offset': 131}, {'char': 'Ρ', 'start_offset': 134, 'end_offset': 136}, {'char': 'Ρ', 'start_offset': 138, 'end_offset': 139}, {'char': 'Π°', 'start_offset': 139, 'end_offset': 140}, {'char': 'Ρ', 'start_offset': 145, 'end_offset': 146}, {'char': 'Π΅', 'start_offset': 146, 'end_offset': 147}, {'char': 'Π³', 'start_offset': 152, 'end_offset': 153}, {'char': 'Ρ', 'start_offset': 153, 'end_offset': 154}, {'char': 'Ρ', 'start_offset': 160, 'end_offset': 161}, {'char': 'Π½', 'start_offset': 167, 'end_offset': 168}, {'char': 'ΠΈ', 'start_offset': 168, 'end_offset': 169}, {'char': 'ΠΉ', 'start_offset': 170, 'end_offset': 171}, {'char': ' ', 'start_offset': 171, 'end_offset': 173}, {'char': 'ΠΏ', 'start_offset': 174, 'end_offset': 175}, {'char': 'Π°', 'start_offset': 176, 'end_offset': 177}, {'char': 'Ρ', 'start_offset': 179, 'end_offset': 180}, {'char': 'Ρ', 'start_offset': 183, 'end_offset': 184}, {'char': 'Π½', 'start_offset': 188, 'end_offset': 189}, {'char': 'Π΅', 'start_offset': 189, 'end_offset': 190}, {'char': 'Ρ', 'start_offset': 193, 'end_offset': 194}, {'char': ' ', 'start_offset': 201, 'end_offset': 203}, {'char': 'ΠΎ', 'start_offset': 204, 'end_offset': 205}, {'char': 'Π΄', 'start_offset': 208, 'end_offset': 209}, {'char': 'Π½', 'start_offset': 214, 'end_offset': 216}, {'char': 'Π°', 'start_offset': 216, 'end_offset': 217}, {'char': 'ΠΊ', 'start_offset': 224, 'end_offset': 225}, {'char': ' ', 'start_offset': 227, 'end_offset': 229}, {'char': 'Ρ', 'start_offset': 233, 'end_offset': 234}, {'char': ' ', 'start_offset': 237, 'end_offset': 239}, {'char': 'Ρ', 'start_offset': 240, 'end_offset': 241}, {'char': 'Ρ', 'start_offset': 241, 'end_offset': 242}, {'char': 'Π·', 'start_offset': 247, 'end_offset': 248}, {'char': 'Π½', 'start_offset': 253, 'end_offset': 254}, {'char': 'ΠΈ', 'start_offset': 254, 'end_offset': 255}, {'char': 'Ρ', 'start_offset': 261, 'end_offset': 262}, {'char': 'Ρ', 'start_offset': 262, 'end_offset': 263}, {'char': ' ', 'start_offset': 281, 'end_offset': 283}, {'char': 'Ρ', 'start_offset': 283, 'end_offset': 284}, {'char': 'Ρ', 'start_offset': 286, 'end_offset': 287}, {'char': 'Π°', 'start_offset': 288, 'end_offset': 289}, {'char': 'Ρ', 'start_offset': 294, 'end_offset': 295}, {'char': 'ΠΈ', 'start_offset': 296, 'end_offset': 297}, {'char': ' ', 'start_offset': 297, 'end_offset': 299}, {'char': 'ΠΌ', 'start_offset': 300, 'end_offset': 301}, {'char': 'Π°', 'start_offset': 301, 'end_offset': 302}, {'char': 'Ρ', 'start_offset': 306, 'end_offset': 307}, {'char': 'Ρ', 'start_offset': 308, 'end_offset': 309}, {'char': 'Ρ', 'start_offset': 309, 'end_offset': 311}, {'char': ' ', 'start_offset': 311, 'end_offset': 313}, {'char': 'Ρ', 'start_offset': 313, 'end_offset': 314}, {'char': 'ΠΏ', 'start_offset': 316, 'end_offset': 317}, {'char': 'Π΅', 'start_offset': 318, 'end_offset': 319}, {'char': 'Ρ', 'start_offset': 324, 'end_offset': 325}, {'char': 'Ρ', 'start_offset': 325, 'end_offset': 326}, {'char': 'Π°', 'start_offset': 328, 'end_offset': 329}, {'char': 'Π»', 'start_offset': 333, 'end_offset': 334}, {'char': 'Ρ', 'start_offset': 334, 'end_offset': 336}, {'char': 'Π½', 'start_offset': 339, 'end_offset': 340}, {'char': 'ΠΈ', 'start_offset': 341, 'end_offset': 342}, {'char': ' ', 'start_offset': 345, 'end_offset': 348}, {'char': 'Π·', 'start_offset': 351, 'end_offset': 352}, {'char': 'Π°', 'start_offset': 354, 'end_offset': 355}, {'char': 'ΠΊ', 'start_offset': 361, 'end_offset': 362}, {'char': 'ΠΎ', 'start_offset': 365, 'end_offset': 366}, {'char': 'Π½', 'start_offset': 373, 'end_offset': 374}, {'char': ' ', 'start_offset': 382, 'end_offset': 384}, {'char': 'Ρ', 'start_offset': 386, 'end_offset': 387}, {'char': 'ΠΊ', 'start_offset': 390, 'end_offset': 391}, {'char': 'ΠΈ', 'start_offset': 392, 'end_offset': 393}, {'char': 'ΠΉ', 'start_offset': 394, 'end_offset': 395}, {'char': ' ', 'start_offset': 396, 'end_offset': 398}, {'char': 'ΠΏ', 'start_offset': 399, 'end_offset': 401}, {'char': 'Π΅', 'start_offset': 402, 'end_offset': 403}, {'char': 'Ρ', 'start_offset': 406, 'end_offset': 407}, {'char': 'Π΅', 'start_offset': 407, 'end_offset': 408}, {'char': 'Π΄', 'start_offset': 411, 'end_offset': 412}, {'char': 'Π±', 'start_offset': 415, 'end_offset': 416}, {'char': 'Π°', 'start_offset': 416, 'end_offset': 417}, {'char': 'Ρ', 'start_offset': 424, 'end_offset': 425}, {'char': 'Π°', 'start_offset': 428, 'end_offset': 429}, {'char': 'Ρ', 'start_offset': 437, 'end_offset': 438}, {'char': ' ', 'start_offset': 445, 'end_offset': 447}, {'char': 'Ρ', 'start_offset': 448, 'end_offset': 449}, {'char': 'ΠΊ', 'start_offset': 452, 'end_offset': 453}, {'char': 'Ρ', 'start_offset': 455, 'end_offset': 456}, {'char': 'ΠΎ', 'start_offset': 457, 'end_offset': 458}, {'char': ' ', 'start_offset': 460, 'end_offset': 463}, {'char': 'ΠΊ', 'start_offset': 463, 'end_offset': 464}, {'char': 'ΠΈ', 'start_offset': 465, 'end_offset': 466}, {'char': 'Ρ', 'start_offset': 470, 'end_offset': 471}, {'char': 'Π°', 'start_offset': 472, 'end_offset': 473}, {'char': 'ΠΉ', 'start_offset': 478, 'end_offset': 480}, {'char': ' ', 'start_offset': 484, 'end_offset': 486}, {'char': 'Π½', 'start_offset': 487, 'end_offset': 488}, {'char': 'Π°', 'start_offset': 488, 'end_offset': 489}, {'char': 'ΠΏ', 'start_offset': 493, 'end_offset': 494}, {'char': 'Π°', 'start_offset': 496, 'end_offset': 497}, {'char': 'Π΄', 'start_offset': 502, 'end_offset': 503}, {'char': 'Π΅', 'start_offset': 504, 'end_offset': 505}, {'char': ' ', 'start_offset': 509, 'end_offset': 511}, {'char': 'Π½', 'start_offset': 511, 'end_offset': 512}, {'char': 'Π°', 'start_offset': 513, 'end_offset': 514}, {'char': ' ', 'start_offset': 515, 'end_offset': 517}, {'char': 'Ρ', 'start_offset': 518, 'end_offset': 519}, {'char': 'Π°', 'start_offset': 519, 'end_offset': 520}, {'char': 'ΠΉ', 'start_offset': 524, 'end_offset': 525}, {'char': 'Π²', 'start_offset': 527, 'end_offset': 528}, {'char': 'Π°', 'start_offset': 529, 'end_offset': 530}, {'char': 'Π½', 'start_offset': 535, 'end_offset': 536}, {'char': 'Ρ', 'start_offset': 536, 'end_offset': 537}, {'char': ' ', 'start_offset': 552, 'end_offset': 555}, {'char': 'Π°', 'start_offset': 555, 'end_offset': 556}, {'char': 'ΠΌ', 'start_offset': 561, 'end_offset': 562}, {'char': 'Π΅', 'start_offset': 562, 'end_offset': 563}, {'char': 'Ρ', 'start_offset': 566, 'end_offset': 567}, {'char': 'ΠΈ', 'start_offset': 567, 'end_offset': 568}, {'char': 'ΠΊ', 'start_offset': 572, 'end_offset': 573}, {'char': 'Π°', 'start_offset': 574, 'end_offset': 575}, {'char': 'Π½', 'start_offset': 579, 'end_offset': 580}, {'char': 'Ρ', 'start_offset': 582, 'end_offset': 583}, {'char': 'Ρ', 'start_offset': 583, 'end_offset': 585}, {'char': 'ΠΊ', 'start_offset': 586, 'end_offset': 587}, {'char': 'ΠΈ', 'start_offset': 588, 'end_offset': 589}, {'char': 'ΠΉ', 'start_offset': 589, 'end_offset': 590}, {'char': ' ', 'start_offset': 591, 'end_offset': 593}, {'char': 'Π²', 'start_offset': 594, 'end_offset': 595}, {'char': 'Ρ', 'start_offset': 595, 'end_offset': 596}, {'char': 'ΠΉ', 'start_offset': 600, 'end_offset': 601}, {'char': 'Ρ', 'start_offset': 604, 'end_offset': 605}, {'char': 'Ρ', 'start_offset': 605, 'end_offset': 607}, {'char': 'ΠΊ', 'start_offset': 609, 'end_offset': 611}, {'char': 'ΠΎ', 'start_offset': 612, 'end_offset': 613}, {'char': 'Π²', 'start_offset': 620, 'end_offset': 621}, {'char': 'Ρ', 'start_offset': 622, 'end_offset': 623}, {'char': ' ', 'start_offset': 637, 'end_offset': 639}, {'char': 'ΠΌ', 'start_offset': 641, 'end_offset': 642}, {'char': 'Π°', 'start_offset': 643, 'end_offset': 644}, {'char': 'Ρ', 'start_offset': 651, 'end_offset': 652}, {'char': 'Ρ', 'start_offset': 654, 'end_offset': 655}, {'char': 'Ρ', 'start_offset': 655, 'end_offset': 656}, {'char': ' ', 'start_offset': 657, 'end_offset': 659}, {'char': 'ΠΉ', 'start_offset': 659, 'end_offset': 660}, {'char': 'ΠΎ', 'start_offset': 660, 'end_offset': 662}, {'char': 'Π³', 'start_offset': 664, 'end_offset': 665}, {'char': 'ΠΎ', 'start_offset': 666, 'end_offset': 667}, {'char': ' ', 'start_offset': 677, 'end_offset': 679}, {'char': 'Π·', 'start_offset': 681, 'end_offset': 682}, {'char': 'Π°', 'start_offset': 683, 'end_offset': 684}, {'char': 'Ρ
', 'start_offset': 686, 'end_offset': 687}, {'char': 'ΠΈ', 'start_offset': 689, 'end_offset': 690}, {'char': 'Ρ', 'start_offset': 696, 'end_offset': 697}, {'char': 'Π°', 'start_offset': 698, 'end_offset': 699}, {'char': 'Ρ', 'start_offset': 707, 'end_offset': 708}, {'char': 'ΠΈ', 'start_offset': 709, 'end_offset': 710}, {'char': ' ', 'start_offset': 733, 'end_offset': 734}, {'char': 'Π΅', 'start_offset': 740, 'end_offset': 741}, {'char': 'Π²', 'start_offset': 747, 'end_offset': 748}, {'char': 'Ρ', 'start_offset': 748, 'end_offset': 749}, {'char': 'ΠΉ', 'start_offset': 752, 'end_offset': 753}, {'char': 'Π²', 'start_offset': 754, 'end_offset': 755}, {'char': 'Π³', 'start_offset': 757, 'end_offset': 758}, {'char': 'Π΅', 'start_offset': 759, 'end_offset': 760}, {'char': 'Ρ', 'start_offset': 767, 'end_offset': 768}, {'char': 'Π΅', 'start_offset': 768, 'end_offset': 769}], word_offsets=[{'word': 'ΠΏΠ°Π½Ρ', 'start_offset': 0, 'end_offset': 12}, {'word': 'ΡΠΏΠΎΠ»ΡΡΠ΅Π½Ρ', 'start_offset': 16, 'end_offset': 39}, {'word': 'ΡΡΠ°ΡΠΈ', 'start_offset': 43, 'end_offset': 59}, {'word': 'Π½Π°Π΄Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ', 'start_offset': 85, 'end_offset': 126}, {'word': 'ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½ΠΈΠΉ', 'start_offset': 130, 'end_offset': 171}, {'word': 'ΠΏΠ°ΡΡΠ½Π΅Ρ', 'start_offset': 174, 'end_offset': 194}, {'word': 'ΠΎΠ΄Π½Π°ΠΊ', 'start_offset': 204, 'end_offset': 225}, {'word': 'Ρ', 'start_offset': 233, 'end_offset': 234}, {'word': 'ΡΡΠ·Π½ΠΈΡΡ', 'start_offset': 240, 'end_offset': 263}, {'word': 'ΡΡΠ°ΡΠΈ', 'start_offset': 283, 'end_offset': 297}, {'word': 'ΠΌΠ°ΡΡΡ', 'start_offset': 300, 'end_offset': 311}, {'word': 'ΡΠΏΠ΅ΡΡΠ°Π»ΡΠ½ΠΈ', 'start_offset': 313, 'end_offset': 342}, {'word': 'Π·Π°ΠΊΠΎΠ½', 'start_offset': 351, 'end_offset': 374}, {'word': 'ΡΠΊΠΈΠΉ', 'start_offset': 386, 'end_offset': 395}, {'word': 'ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ°Ρ', 'start_offset': 399, 'end_offset': 438}, {'word': 'ΡΠΊΡΠΎ', 'start_offset': 448, 'end_offset': 458}, {'word': 'ΠΊΠΈΡΠ°ΠΉ', 'start_offset': 463, 'end_offset': 480}, {'word': 'Π½Π°ΠΏΠ°Π΄Π΅', 'start_offset': 487, 'end_offset': 505}, {'word': 'Π½Π°', 'start_offset': 511, 'end_offset': 514}, {'word': 'ΡΠ°ΠΉΠ²Π°Π½Ρ', 'start_offset': 518, 'end_offset': 537}, {'word': 'Π°ΠΌΠ΅ΡΠΈΠΊΠ°Π½ΡΡΠΊΠΈΠΉ', 'start_offset': 555, 'end_offset': 590}, {'word': 'Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ', 'start_offset': 594, 'end_offset': 623}, {'word': 'ΠΌΠ°ΡΡΡ', 'start_offset': 641, 'end_offset': 656}, {'word': 'ΠΉΠΎΠ³ΠΎ', 'start_offset': 659, 'end_offset': 667}, {'word': 'Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ', 'start_offset': 681, 'end_offset': 710}, {'word': 'Π΅Π²ΡΠΉΠ²Π³Π΅ΡΠ΅', 'start_offset': 740, 'end_offset': 769}])
|
153 |
-
```
|
154 |
-
|
155 |
-
### Split by seconds
|
156 |
-
|
157 |
-
```
|
158 |
-
0.0 - 0.24: ΠΏΠ°Π½Ρ
|
159 |
-
0.32 - 0.78: ΡΠΏΠΎΠ»ΡΡΠ΅Π½Ρ
|
160 |
-
0.86 - 1.18: ΡΡΠ°ΡΠΈ
|
161 |
-
1.7 - 2.52: Π½Π°Π΄Π²Π°ΠΆΠ»ΠΈΠ²ΠΈΠΉ
|
162 |
-
2.6 - 3.42: ΡΡΡΠ°ΡΠ΅Π³ΡΡΠ½ΠΈΠΉ
|
163 |
-
3.48 - 3.88: ΠΏΠ°ΡΡΠ½Π΅Ρ
|
164 |
-
4.08 - 4.5: ΠΎΠ΄Π½Π°ΠΊ
|
165 |
-
4.66 - 4.68: Ρ
|
166 |
-
4.8 - 5.26: ΡΡΠ·Π½ΠΈΡΡ
|
167 |
-
5.66 - 5.94: ΡΡΠ°ΡΠΈ
|
168 |
-
6.0 - 6.22: ΠΌΠ°ΡΡΡ
|
169 |
-
6.26 - 6.84: ΡΠΏΠ΅ΡΡΠ°Π»ΡΠ½ΠΈ
|
170 |
-
7.02 - 7.48: Π·Π°ΠΊΠΎΠ½
|
171 |
-
7.72 - 7.9: ΡΠΊΠΈΠΉ
|
172 |
-
7.98 - 8.76: ΠΏΠ΅ΡΠ΅Π΄Π±Π°ΡΠ°Ρ
|
173 |
-
8.96 - 9.16: ΡΠΊΡΠΎ
|
174 |
-
9.26 - 9.6: ΠΊΠΈΡΠ°ΠΉ
|
175 |
-
9.74 - 10.1: Π½Π°ΠΏΠ°Π΄Π΅
|
176 |
-
10.22 - 10.28: Π½Π°
|
177 |
-
10.36 - 10.74: ΡΠ°ΠΉΠ²Π°Π½Ρ
|
178 |
-
11.1 - 11.8: Π°ΠΌΠ΅ΡΠΈΠΊΠ°Π½ΡΡΠΊΠΈΠΉ
|
179 |
-
11.88 - 12.46: Π²ΡΠΉΡΡΠΊΠΎΠ²Ρ
|
180 |
-
12.82 - 13.12: ΠΌΠ°ΡΡΡ
|
181 |
-
13.18 - 13.34: ΠΉΠΎΠ³ΠΎ
|
182 |
-
13.62 - 14.2: Π·Π°Ρ
ΠΈΡΠ°ΡΠΈ
|
183 |
-
14.8 - 15.38: Π΅Π²ΡΠΉΠ²Π³Π΅ΡΠ΅
|
184 |
```
|
|
|
1 |
+
# `gradio-learning`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
+
## Install
|
4 |
|
5 |
+
```shell
|
6 |
+
uv venv --python 3.12
|
7 |
|
8 |
+
source .venv/bin/activate
|
|
|
9 |
|
10 |
+
uv pip install -r requirements.txt
|
11 |
|
12 |
+
# in development mode
|
13 |
+
uv pip install -r requirements-dev.txt
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
```
|
app.py
CHANGED
@@ -1,93 +1,132 @@
|
|
1 |
-
import
|
2 |
-
|
3 |
import torch
|
4 |
-
import
|
5 |
|
6 |
-
|
7 |
|
8 |
-
|
9 |
-
processor = Wav2Vec2ProcessorWithLM.from_pretrained(model_name)
|
10 |
-
model = Wav2Vec2ForCTC.from_pretrained(model_name)
|
11 |
-
model.to("cpu")
|
12 |
|
|
|
|
|
|
|
13 |
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
if max_seconds > 0:
|
25 |
-
speech_array = speech_array[: max_seconds * 16000]
|
26 |
-
batch["speech"] = speech_array.numpy()
|
27 |
-
batch["sampling_rate"] = 16000
|
28 |
-
return batch
|
29 |
-
|
30 |
-
|
31 |
-
# tokenize
|
32 |
-
def inference(audio):
|
33 |
-
# read in sound file
|
34 |
-
# load dummy dataset and read soundfiles
|
35 |
-
sp = speech_file_to_array_fn(audio.name)
|
36 |
-
|
37 |
-
sample_rate = 16000
|
38 |
-
# stride_length_s is a tuple of the left and right stride length.
|
39 |
-
# With only 1 number, both sides get the same stride, by default
|
40 |
-
# the stride_length on one side is 1/6th of the chunk_length_s
|
41 |
-
input_values = processor(
|
42 |
-
sp["speech"],
|
43 |
-
sample_rate=sample_rate,
|
44 |
-
chunk_length_s=10,
|
45 |
-
stride_length_s=(4, 2),
|
46 |
-
return_tensors="pt",
|
47 |
-
).input_values
|
48 |
-
|
49 |
-
with torch.no_grad():
|
50 |
-
logits = model(input_values).logits
|
51 |
-
|
52 |
-
pred_ids = torch.argmax(logits, axis=-1).cpu().tolist()
|
53 |
-
prediction = tokenizer.decode(pred_ids[0], output_word_offsets=True)
|
54 |
-
|
55 |
-
time_offset = 320 / sample_rate
|
56 |
-
|
57 |
-
total_prediction = []
|
58 |
-
words = []
|
59 |
-
for item in prediction.word_offsets:
|
60 |
-
r = item
|
61 |
-
|
62 |
-
s = round(r['start_offset'] * time_offset, 2)
|
63 |
-
e = round(r['end_offset'] * time_offset, 2)
|
64 |
-
|
65 |
-
total_prediction.append(f"{s} - {e}: {r['word']}")
|
66 |
-
words.append(r['word'])
|
67 |
-
|
68 |
-
print(prediction[0])
|
69 |
-
|
70 |
-
return "\n".join(total_prediction) + "\n\n" + ' '.join(words)
|
71 |
-
|
72 |
-
|
73 |
-
inputs = gr.inputs.Audio(label="Input Audio", type="file")
|
74 |
-
outputs = gr.outputs.Textbox(label="Output Text")
|
75 |
-
title = model_name
|
76 |
-
description = f"Gradio demo for a {model_name}. To use it, simply upload your audio, or click one of the examples to load them. Read more at the links below. Currently supports .wav 16_000hz files"
|
77 |
-
article = "<p style='text-align: center'><a href='https://github.com/egorsmkv/wav2vec2-uk-demo' target='_blank'> Github repo</a> | <a href='<HF Space link>' target='_blank'>Pretrained model</a> | Made with help from <a href='https://github.com/robinhad' target='_blank'>@robinhad</a></p>"
|
78 |
-
examples = [
|
79 |
-
["long_1.wav"],
|
80 |
-
["mer_lviv_interview.wav"],
|
81 |
-
["short_1.wav"],
|
82 |
-
["tsn_2.wav"],
|
83 |
-
["tsn.wav"],
|
84 |
]
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import time
|
2 |
+
|
3 |
import torch
|
4 |
+
import librosa
|
5 |
|
6 |
+
import gradio as gr
|
7 |
|
8 |
+
from transformers import AutoModelForCTC, Wav2Vec2BertProcessor
|
|
|
|
|
|
|
9 |
|
10 |
+
model_name = "Yehor/w2v-bert-2.0-uk"
|
11 |
+
device = "cpu"
|
12 |
+
max_duration = 30
|
13 |
|
14 |
+
asr_model = AutoModelForCTC.from_pretrained(model_name).to(device)
|
15 |
+
processor = Wav2Vec2BertProcessor.from_pretrained(model_name)
|
16 |
+
|
17 |
+
audio_samples = [
|
18 |
+
"sample_1.wav",
|
19 |
+
"sample_2.wav",
|
20 |
+
"sample_3.wav",
|
21 |
+
"sample_4.wav",
|
22 |
+
"sample_5.wav",
|
23 |
+
"sample_6.wav",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
]
|
25 |
+
|
26 |
+
description_head = """
|
27 |
+
# Speech-to-Text for Ukrainian
|
28 |
+
|
29 |
+
## Overview
|
30 |
+
|
31 |
+
This space uses https://huggingface.co/Yehor/w2v-bert-2.0-uk model that solves
|
32 |
+
a Speech-to-Text task for the Ukrainian language.
|
33 |
+
""".strip()
|
34 |
+
|
35 |
+
description_foot = """
|
36 |
+
## Community
|
37 |
+
|
38 |
+
- Join our Discord server - https://discord.gg/yVAjkBgmt4 - where we're talking about Data Science,
|
39 |
+
Machine Learning, Deep Learning, and Artificial Intelligence.
|
40 |
+
|
41 |
+
- Join our Speech Recognition Group in Telegram: https://t.me/speech_recognition_uk
|
42 |
+
""".strip()
|
43 |
+
|
44 |
+
|
45 |
+
def inference(audio_path, progress=gr.Progress()):
|
46 |
+
gr.Info("Starting process", duration=2)
|
47 |
+
|
48 |
+
progress(0, desc="Starting")
|
49 |
+
|
50 |
+
duration = librosa.get_duration(path=audio_path)
|
51 |
+
if duration > max_duration:
|
52 |
+
raise gr.Error("The duration of the file exceeds 10 seconds.")
|
53 |
+
|
54 |
+
paths = [
|
55 |
+
audio_path,
|
56 |
+
]
|
57 |
+
|
58 |
+
results = []
|
59 |
+
|
60 |
+
for path in progress.tqdm(paths, desc="Recognizing...", unit="file"):
|
61 |
+
t0 = time.time()
|
62 |
+
|
63 |
+
audio_duration = librosa.get_duration(path=path, sr=16_000)
|
64 |
+
audio_input, _ = librosa.load(path, mono=True, sr=16_000)
|
65 |
+
|
66 |
+
features = processor([audio_input], sampling_rate=16_000).input_features
|
67 |
+
features = torch.tensor(features).to(device)
|
68 |
+
|
69 |
+
with torch.inference_mode():
|
70 |
+
logits = asr_model(features).logits
|
71 |
+
|
72 |
+
predicted_ids = torch.argmax(logits, dim=-1)
|
73 |
+
predictions = processor.batch_decode(predicted_ids)
|
74 |
+
|
75 |
+
elapsed_time = round(time.time() - t0, 2)
|
76 |
+
rtf = round(elapsed_time / audio_duration, 4)
|
77 |
+
audio_duration = round(audio_duration, 2)
|
78 |
+
|
79 |
+
results.append(
|
80 |
+
{
|
81 |
+
"path": path.split("/")[-1],
|
82 |
+
"transcription": "\n".join(predictions),
|
83 |
+
"audio_duration": audio_duration,
|
84 |
+
"rtf": rtf,
|
85 |
+
}
|
86 |
+
)
|
87 |
+
|
88 |
+
gr.Info("Finished...", duration=2)
|
89 |
+
|
90 |
+
result_texts = []
|
91 |
+
|
92 |
+
for result in results:
|
93 |
+
result_texts.append(f'**{result["path"]}**')
|
94 |
+
result_texts.append("\n\n")
|
95 |
+
result_texts.append(f"> {result['transcription']}")
|
96 |
+
result_texts.append("\n\n")
|
97 |
+
result_texts.append(f'**Audio duration**: {result['audio_duration']}')
|
98 |
+
result_texts.append("\n")
|
99 |
+
result_texts.append(f'**Real-Time Factor**: {result['rtf']}')
|
100 |
+
|
101 |
+
return "\n".join(result_texts)
|
102 |
+
|
103 |
+
|
104 |
+
demo = gr.Blocks(
|
105 |
+
title="Speech-to-Text for Ukrainian",
|
106 |
+
analytics_enabled=False,
|
107 |
+
)
|
108 |
+
|
109 |
+
with demo:
|
110 |
+
gr.Markdown(description_head)
|
111 |
+
|
112 |
+
gr.Markdown(f"## Demo (max. duration: **{max_duration}** seconds)")
|
113 |
+
|
114 |
+
with gr.Row():
|
115 |
+
audio_file = gr.Audio(label="Audio file", type="filepath")
|
116 |
+
transcription = gr.Markdown(
|
117 |
+
label="Transcription",
|
118 |
+
value="Recognized text will appear here. Use **an example file** below the Recognize button,"
|
119 |
+
"upload **your audio file**, or use **the microphone** to record something...",
|
120 |
+
)
|
121 |
+
|
122 |
+
gr.Button("Recognize").click(inference, inputs=audio_file, outputs=transcription)
|
123 |
+
|
124 |
+
with gr.Row():
|
125 |
+
gr.Examples(
|
126 |
+
label="Choose an example audio", inputs=audio_file, examples=audio_samples
|
127 |
+
)
|
128 |
+
|
129 |
+
gr.Markdown(description_foot)
|
130 |
+
|
131 |
+
if __name__ == "__main__":
|
132 |
+
demo.launch()
|
inference.py
DELETED
@@ -1,68 +0,0 @@
|
|
1 |
-
import argparse
|
2 |
-
import torch
|
3 |
-
import torchaudio
|
4 |
-
from pathlib import Path
|
5 |
-
from transformers import Wav2Vec2ProcessorWithLM, Wav2Vec2ForCTC
|
6 |
-
|
7 |
-
|
8 |
-
def main(args):
|
9 |
-
processor = Wav2Vec2ProcessorWithLM.from_pretrained(args.model_id)
|
10 |
-
model = Wav2Vec2ForCTC.from_pretrained(args.model_id)
|
11 |
-
model.to('cpu')
|
12 |
-
|
13 |
-
files = args.path_files.split(',')
|
14 |
-
|
15 |
-
for path_file in files:
|
16 |
-
print('File:', path_file)
|
17 |
-
|
18 |
-
wav_file_path = str(Path(path_file).absolute())
|
19 |
-
waveform, sample_rate = torchaudio.load(wav_file_path)
|
20 |
-
|
21 |
-
if sample_rate != 16000:
|
22 |
-
resample = torchaudio.transforms.Resample(
|
23 |
-
sample_rate, 16000, resampling_method='sinc_interpolation')
|
24 |
-
speech_array = resample(waveform)
|
25 |
-
sp = speech_array.squeeze().numpy()
|
26 |
-
else:
|
27 |
-
sp = waveform.squeeze().numpy()
|
28 |
-
|
29 |
-
# stride_length_s is a tuple of the left and right stride length.
|
30 |
-
# With only 1 number, both sides get the same stride, by default
|
31 |
-
# the stride_length on one side is 1/6th of the chunk_length_s
|
32 |
-
input_values = processor(sp,
|
33 |
-
sample_rate=16000,
|
34 |
-
chunk_length_s=args.chunk_length_s,
|
35 |
-
stride_length_s=(args.stride_length_s_l, args.stride_length_s_r),
|
36 |
-
return_tensors="pt").input_values
|
37 |
-
|
38 |
-
with torch.no_grad():
|
39 |
-
logits = model(input_values).logits
|
40 |
-
|
41 |
-
prediction = processor.batch_decode(logits.numpy()).text
|
42 |
-
print(prediction[0])
|
43 |
-
|
44 |
-
|
45 |
-
if __name__ == "__main__":
|
46 |
-
parser = argparse.ArgumentParser()
|
47 |
-
|
48 |
-
parser.add_argument(
|
49 |
-
"--path_files", type=str, required=True, help="WAV files to transcribe, separated by a comma"
|
50 |
-
)
|
51 |
-
parser.add_argument(
|
52 |
-
"--model_id", type=str, required=True, help="Model identifier. Should be loadable with π€ Transformers"
|
53 |
-
)
|
54 |
-
parser.add_argument(
|
55 |
-
"--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
|
56 |
-
)
|
57 |
-
parser.add_argument(
|
58 |
-
"--stride_length_s_l", type=int, default=None, help="Stride of the audio chunks, left value."
|
59 |
-
)
|
60 |
-
parser.add_argument(
|
61 |
-
"--stride_length_s_r", type=int, default=None, help="Stride of the audio chunks, right value."
|
62 |
-
)
|
63 |
-
parser.add_argument(
|
64 |
-
"--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
|
65 |
-
)
|
66 |
-
args = parser.parse_args()
|
67 |
-
|
68 |
-
main(args)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
inference_gpu.py
DELETED
@@ -1,69 +0,0 @@
|
|
1 |
-
import argparse
|
2 |
-
import torch
|
3 |
-
import torchaudio
|
4 |
-
from pathlib import Path
|
5 |
-
from transformers import Wav2Vec2ProcessorWithLM, Wav2Vec2ForCTC
|
6 |
-
|
7 |
-
|
8 |
-
def main(args):
|
9 |
-
processor = Wav2Vec2ProcessorWithLM.from_pretrained(args.model_id)
|
10 |
-
model = Wav2Vec2ForCTC.from_pretrained(args.model_id)
|
11 |
-
model.to('cuda')
|
12 |
-
|
13 |
-
files = args.path_files.split(',')
|
14 |
-
|
15 |
-
for path_file in files:
|
16 |
-
print('File:', path_file)
|
17 |
-
|
18 |
-
wav_file_path = str(Path(path_file).absolute())
|
19 |
-
waveform, sample_rate = torchaudio.load(wav_file_path)
|
20 |
-
|
21 |
-
if sample_rate != 16000:
|
22 |
-
resample = torchaudio.transforms.Resample(
|
23 |
-
sample_rate, 16000, resampling_method='sinc_interpolation')
|
24 |
-
speech_array = resample(waveform)
|
25 |
-
sp = speech_array.squeeze().numpy()
|
26 |
-
else:
|
27 |
-
sp = waveform.squeeze().numpy()
|
28 |
-
|
29 |
-
# stride_length_s is a tuple of the left and right stride length.
|
30 |
-
# With only 1 number, both sides get the same stride, by default
|
31 |
-
# the stride_length on one side is 1/6th of the chunk_length_s
|
32 |
-
input_values = processor(sp,
|
33 |
-
sample_rate=16000,
|
34 |
-
chunk_length_s=args.chunk_length_s,
|
35 |
-
stride_length_s=(args.stride_length_s_l, args.stride_length_s_r),
|
36 |
-
return_tensors="pt").input_values
|
37 |
-
input_values = input_values.cuda()
|
38 |
-
|
39 |
-
with torch.no_grad():
|
40 |
-
logits = model(input_values).logits
|
41 |
-
|
42 |
-
prediction = processor.batch_decode(logits.cpu().numpy()).text
|
43 |
-
print(prediction[0])
|
44 |
-
|
45 |
-
|
46 |
-
if __name__ == "__main__":
|
47 |
-
parser = argparse.ArgumentParser()
|
48 |
-
|
49 |
-
parser.add_argument(
|
50 |
-
"--path_files", type=str, required=True, help="WAV files to transcribe, separated by a comma"
|
51 |
-
)
|
52 |
-
parser.add_argument(
|
53 |
-
"--model_id", type=str, required=True, help="Model identifier. Should be loadable with π€ Transformers"
|
54 |
-
)
|
55 |
-
parser.add_argument(
|
56 |
-
"--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
|
57 |
-
)
|
58 |
-
parser.add_argument(
|
59 |
-
"--stride_length_s_l", type=int, default=None, help="Stride of the audio chunks, left value."
|
60 |
-
)
|
61 |
-
parser.add_argument(
|
62 |
-
"--stride_length_s_r", type=int, default=None, help="Stride of the audio chunks, right value."
|
63 |
-
)
|
64 |
-
parser.add_argument(
|
65 |
-
"--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
|
66 |
-
)
|
67 |
-
args = parser.parse_args()
|
68 |
-
|
69 |
-
main(args)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
inference_timestamps.py
DELETED
@@ -1,86 +0,0 @@
|
|
1 |
-
import argparse
|
2 |
-
from time import gmtime, strftime
|
3 |
-
|
4 |
-
import torch
|
5 |
-
import torchaudio
|
6 |
-
from pathlib import Path
|
7 |
-
from transformers import Wav2Vec2ProcessorWithLM, Wav2Vec2ForCTC, Wav2Vec2CTCTokenizer
|
8 |
-
|
9 |
-
|
10 |
-
def main(args):
|
11 |
-
tokenizer = Wav2Vec2CTCTokenizer.from_pretrained(args.model_id)
|
12 |
-
processor = Wav2Vec2ProcessorWithLM.from_pretrained(args.model_id)
|
13 |
-
model = Wav2Vec2ForCTC.from_pretrained(args.model_id)
|
14 |
-
model.to('cpu')
|
15 |
-
|
16 |
-
files = args.path_files.split(',')
|
17 |
-
|
18 |
-
for path_file in files:
|
19 |
-
print('File:', path_file)
|
20 |
-
|
21 |
-
wav_file_path = str(Path(path_file).absolute())
|
22 |
-
waveform, sample_rate = torchaudio.load(wav_file_path)
|
23 |
-
|
24 |
-
if sample_rate != 16000:
|
25 |
-
resample = torchaudio.transforms.Resample(
|
26 |
-
sample_rate, 16000, resampling_method='sinc_interpolation')
|
27 |
-
sample_rate = 16000
|
28 |
-
speech_array = resample(waveform)
|
29 |
-
sp = speech_array.squeeze().numpy()
|
30 |
-
else:
|
31 |
-
sp = waveform.squeeze().numpy()
|
32 |
-
|
33 |
-
# stride_length_s is a tuple of the left and right stride length.
|
34 |
-
# With only 1 number, both sides get the same stride, by default
|
35 |
-
# the stride_length on one side is 1/6th of the chunk_length_s
|
36 |
-
input_values = processor(sp,
|
37 |
-
sample_rate=16000,
|
38 |
-
chunk_length_s=args.chunk_length_s,
|
39 |
-
stride_length_s=(args.stride_length_s_l, args.stride_length_s_r),
|
40 |
-
return_tensors="pt").input_values
|
41 |
-
|
42 |
-
with torch.no_grad():
|
43 |
-
logits = model(input_values).logits
|
44 |
-
|
45 |
-
# prediction = tokenizer.decode(pred_ids[0], output_word_offsets=True)
|
46 |
-
# prediction = tokenizer.decode(pred_ids[0], output_char_offsets=True)
|
47 |
-
|
48 |
-
pred_ids = torch.argmax(logits, axis=-1).cpu().tolist()
|
49 |
-
prediction = tokenizer.decode(pred_ids[0], output_word_offsets=True)
|
50 |
-
|
51 |
-
print(f'Sample rate: {sample_rate}')
|
52 |
-
time_offset = 320 / sample_rate
|
53 |
-
|
54 |
-
for item in prediction.word_offsets:
|
55 |
-
r = item
|
56 |
-
|
57 |
-
s = round(r['start_offset'] * time_offset, 2)
|
58 |
-
e = round(r['end_offset'] * time_offset, 2)
|
59 |
-
|
60 |
-
print(f"{s} - {e}: {r['word']}")
|
61 |
-
|
62 |
-
|
63 |
-
if __name__ == "__main__":
|
64 |
-
parser = argparse.ArgumentParser()
|
65 |
-
|
66 |
-
parser.add_argument(
|
67 |
-
"--path_files", type=str, required=True, help="WAV files to transcribe, separated by a comma"
|
68 |
-
)
|
69 |
-
parser.add_argument(
|
70 |
-
"--model_id", type=str, required=True, help="Model identifier. Should be loadable with π€ Transformers"
|
71 |
-
)
|
72 |
-
parser.add_argument(
|
73 |
-
"--chunk_length_s", type=float, default=None, help="Chunk length in seconds. Defaults to 5 seconds."
|
74 |
-
)
|
75 |
-
parser.add_argument(
|
76 |
-
"--stride_length_s_l", type=int, default=None, help="Stride of the audio chunks, left value."
|
77 |
-
)
|
78 |
-
parser.add_argument(
|
79 |
-
"--stride_length_s_r", type=int, default=None, help="Stride of the audio chunks, right value."
|
80 |
-
)
|
81 |
-
parser.add_argument(
|
82 |
-
"--log_outputs", action="store_true", help="If defined, write outputs to log file for analysis."
|
83 |
-
)
|
84 |
-
args = parser.parse_args()
|
85 |
-
|
86 |
-
main(args)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mer_lviv_interview.wav
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:921bda3185b93787b15887a715688801a4fa9c2a0c255cfbdc674380a21f9d17
|
3 |
-
size 12759536
|
|
|
|
|
|
|
|
requirements-dev.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
ruff
|
requirements.txt
CHANGED
@@ -1,9 +1,8 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
torchaudio===0.10.2
|
|
|
1 |
+
gradio
|
2 |
+
|
3 |
+
torch
|
4 |
+
torchaudio
|
5 |
+
|
6 |
+
transformers
|
7 |
+
|
8 |
+
librosa
|
|
tsn_2.wav β sample_1.wav
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:172ade978b299f4a0c47e3b76666d1a06161e6001fbb5591b82038a1bbc4b5ad
|
3 |
+
size 272568
|
short_1.wav β sample_2.wav
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:98fe42f22f8ea632714081a958dc035f3d507523fd340b320a1223ac2f55ccac
|
3 |
+
size 199942
|
tsn.wav β sample_3.wav
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:83c0b7375beada8cee74b5de226da494368fcc6a3ce692913b3302dcda0bd9a2
|
3 |
+
size 192842
|
long_1.wav β sample_4.wav
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:19e466ee9c0c129c1eecf93eb6791a44c2ee8d68dce2c3e8fd3734b87f28324a
|
3 |
+
size 241442
|
sample_5.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5af19120c92859846a08496e0a617c21877cae2db5807d211f0a431d95163a3e
|
3 |
+
size 193388
|
sample_6.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ac877968d5749438930339497f7548046003390a848496136f6cbe8a74c51629
|
3 |
+
size 186290
|