1 00:35:32,779 --> 00:35:36,779 OK Google 2 00:35:38,054 --> 00:35:42,054 hello test 3 00:35:50,839 --> 00:35:54,839 hello test 4 00:36:24,407 --> 00:36:28,407 good morning welcome to Thailand captioning service center 5 00:36:30,516 --> 00:36:34,516 as some of you may already know the captioning service exception is the service 6 00:36:40,956 --> 00:36:44,956 that help people with hearing disability to understand the audio content by reading the transcription instead of listen to the audio 7 00:36:51,330 --> 00:36:55,330 so as I did send her before I brought them already happening and also the to record caption 8 00:37:02,034 --> 00:37:03,300 I thought if you record content and you see on the screen here are they more of the live caption is it the caption of what I'm thinking right now 9 00:37:03,300 --> 00:37:07,300 so 10 00:37:10,840 --> 00:37:14,840 at this center we use three different techniques to Hawaii we don't have a chimney what is a challenging 11 00:37:16,656 --> 00:37:20,656 the real competition is that you have to provide a check tax return instantly 12 00:37:21,848 --> 00:37:25,848 within 5 or 10 second after the speaker is speaking. 13 00:37:31,949 --> 00:37:35,949 and also you have to provide the corrected text more than ninety percent accuracy so that the people will read that has Christian Canada 14 00:37:40,750 --> 00:37:44,750 we provide the service here does Sobeys grid centers in our 2017 15 00:37:51,743 --> 00:37:55,324 actually we are specialized with live captions and used to different technique to do the Raton Pass Christian 16 00:37:55,324 --> 00:37:59,324 how to treat headaches sbus 17 00:38:02,020 --> 00:38:06,020 what kind of typing revised and automatic speech recognition system 18 00:38:06,764 --> 00:38:09,979 we have to use two different techniques to make is suitable for 19 00:38:09,979 --> 00:38:13,979 difference speaking a and other 20 00:38:15,008 --> 00:38:19,008 quality of inputs of speech 21 00:39:18,343 --> 00:39:22,343 test 1 2 3 4 22 00:41:02,943 --> 00:41:06,247 okay 23 00:41:06,247 --> 00:41:10,247 if anyone is here 24 00:44:33,087 --> 00:44:37,087 hello Test 25 00:44:55,545 --> 00:44:59,545 Good morning . Welcome to Thailand captioning service center 26 00:45:07,835 --> 00:45:11,835 as some of you may already know the captaining service is the service that help people with hearing disability understand audio content and your content by 27 00:45:12,240 --> 00:45:16,240 by reading that the text instead of listening to the audio 28 00:45:17,742 --> 00:45:21,742 here at the center people wide attaching themselves for both 29 00:45:24,604 --> 00:45:28,604 Real-time captions or li caption is used for Life program 30 00:45:31,833 --> 00:45:35,833 and also the caption for recording content 31 00:45:39,052 --> 00:45:43,052 . At the screen you see the demo of the real ham has suction I will try captain 32 00:45:43,344 --> 00:45:47,344 This is of the captioning on what I'm currently speaking 33 00:45:48,951 --> 00:45:52,951 a descender we start in 34 00:45:53,862 --> 00:45:54,687 2017 and we use three different techniques 35 00:45:54,687 --> 00:45:58,687 to 36 00:46:45,953 --> 00:46:49,953 37 00:47:08,644 --> 00:47:12,644 at this sender we provide captoin service 38 00:47:14,377 --> 00:47:17,565 time caption and pre-record content in is used 39 00:47:17,565 --> 00:47:21,212 spider costume as simina and 40 00:47:21,212 --> 00:47:22,978 television programs flyer program Chavez Paso news 41 00:47:22,978 --> 00:47:26,978 as you seen on this screen 42 00:47:29,169 --> 00:47:33,169 you see the last caption of what I'm currently speaking 43 00:47:33,249 --> 00:47:35,388 . what is challenging with the real captioning is that 44 00:47:35,388 --> 00:47:39,388 you have to provide the captions 45 00:47:43,809 --> 00:47:47,809 instantly within 5 to 10 second after the speaker is speaking and you also have to provide to kept on 46 00:47:51,511 --> 00:47:55,511 accurate enough for people who read the caption to understand the text the should a procedure should be higher than 90% 47 00:47:59,299 --> 00:48:02,677 at our Center we use two different techniques to do the real ham caption 48 00:48:02,677 --> 00:48:05,943 three different techniques are simontaneous highness typing 49 00:48:05,943 --> 00:48:09,943 why's that automatically commission s recognition 50 00:48:15,995 --> 00:48:19,995 used three different technique because you select the one that is suitable for the speaking style and the quality of speech 51 00:48:24,234 --> 00:48:26,906 the first one we called a typing we use this technique when we first got our Center 52 00:48:26,906 --> 00:48:30,906 Fox Southeast yato 53 00:48:38,814 --> 00:48:42,814 because at that time to automatic speech recognition but high is not accurate in out the accuracy around 70 to 80% 54 00:48:46,892 --> 00:48:50,892 do we have to use the typist to type finally and we use treatable type is 2 time 55 00:48:52,713 --> 00:48:55,442 together in manner to provide a caption in your 56 00:48:55,442 --> 00:48:57,676 and you will see the type is ended 57 00:48:57,676 --> 00:49:01,676 58 00:49:06,817 --> 00:49:09,691 this room to do it as somewhat helping what is the real time catching for costume in I and 59 00:49:09,691 --> 00:49:13,691 you see this 60 00:49:16,004 --> 00:49:17,877 I thought you took them your type to text transcription up about 61 00:49:17,877 --> 00:49:21,877 5 Second of speech 62 00:49:22,705 --> 00:49:25,908 and then we have also added the fifth person doing the 63 00:49:25,908 --> 00:49:29,503 text to correct that takes it be better 64 00:49:29,503 --> 00:49:30,772 on the left hand side 65 00:49:30,772 --> 00:49:32,551 66 00:49:32,551 --> 00:49:36,551 the typist 67 00:49:37,071 --> 00:49:41,071 who do the corrected of the 68 00:49:43,872 --> 00:49:47,872 speech recognition this is what we use for the real Krampus ending of this demo 69 00:49:51,388 --> 00:49:55,388 to edit a how is ato correct the music recognition from the automatic speech recognition 70 00:49:58,322 --> 00:50:02,053 add me move on to the next room the last technic this video 71 00:50:02,053 --> 00:50:06,053 call 72 00:50:06,197 --> 00:50:10,197 Re-voice so you have a studio here I should leave you have 73 00:50:15,329 --> 00:50:17,791 Studio Re-voice that we have an engine 74 00:50:17,791 --> 00:50:18,486 75 00:50:18,486 --> 00:50:21,525 76 00:50:21,525 --> 00:50:24,043 original speaker and then 77 00:50:24,043 --> 00:50:28,043 speak that are try again to get in the 78 00:50:30,948 --> 00:50:34,948 clear Manor in the quiet room to improve the quality of the speech. 79 00:50:44,677 --> 00:50:46,501 you can also try it by yourself for the first of all 80 00:50:46,501 --> 00:50:50,501 we 81 00:50:50,669 --> 00:50:54,669 should login in the here 82 00:51:05,987 --> 00:51:09,987 cut okay