 Where you can directly copy those files into the server and then simply put all his data using Excel sheet. How exactly the metadata is created in the SRT? The SRT and the text generation, the SRT is generated for videos and the text is generated for PDF and PPT. How the metadata is collected? The SRT is generated using a software named as Autosub. So where from the metadata will come? As soon as we upload one video. That is all non-technical answer. I am asking a technical question now. Suppose I have uploaded a video. Let's take Sachin Tendulkar completing his 10,000 runs. That clip I have uploaded. How come the search engine will come to know that he is a Sachin Tendulkar. He has completed 1,000 runs. This is the particular date on he has completed. We couldn't implement that because it needs hierarchical data structure. We are using simple data structure. Suppose somebody types Sachin, then what Google does is it will show that he has recently completed 10,000 runs. Where from the information will come? My question is that this is called what we call? All this information will come from SRT subtitles only. Subtitles. Subtitles and the names of the files. Who will put these subtitles now? Just when we upload the video, all the subtitles will be generated automatically in the backend. Who will generate subtitles? Our code will generate automatically. What kind of subtitles can you define me? Subtitles. We have used one software named as Autosub. Once you upload the file, the Autosub, the R software will convert all the audio to text conversion in any language. It uses Google's text to speech. Whether it works? Yes, it works. We have implemented some videos on Fataxers and other persons who actually conducted those workshops. We will show that in the demo. All the subtitles. I don't know the demo. Okay, so we'll directly go for the demo. So here, this is the particular content management page. The administrator has to choose one course. So if he chooses, suppose computer programming, then he gets all the files in sorted manner as PDF, PPD or video. Suppose he wants to see the subtitle of PDF. So when he clicks on the review, this is the particular subtitle generated for this particular video. Which is the subtitle? This is the flowing subtitle. If you start the video, subtitles will go up automatically. Even if you click on this particular line, you can directly come to that particular point of the time. And even if you drag this particular video forward, the subtitle will go forward. So this is based on audio, correct? Yes. So suppose the audio doesn't many times replicate what exactly the video is for. So then in that case, what will happen? This particular interface is given for administrators. So he can edit the subtitles. It's okay. But suppose what I'm saying that. Let's take the first example. Sajith Tendulkar hit in 1000 runs. Yes. So the commentator will say all other things rather than the, suppose the main statement that he has completed 1000 runs. So that main important metadata will not come. Correct? So it must be present in the title of the video, right? Where form the sourcing will happen? Machine, how come understand? Machine. Machine in the sense PC. In search. Not search, creating the metadata. So the source of metadata is audio? Yes. Suppose in audio, the particular statements are not available will not become a part of search engine, correct? Yes. They won't become part of search engine. Meaning thereby it is an audio based search or commentary search, not video search. Correct? Yes, but generally what we are implementing is we are implementing the lecture series, right? Suppose in this particular lecture, Phatak Sir is teaching about computer programming. So whatever he talks about computer programming should appear in the search result, right? What I'm saying that it is the commentary search rather than the video search. Is it okay? Yes. So what you are claiming initially that is a video. No, but in general the audio is must be related, generally it is related to video. Not necessarily. I may have got a clip on which there is not a single statement only the instrumental music is going on, complete documentary. Quite possible. There are calligraphic animation in which only the words are coming. You go to the TV channel EPIC, you will find so many documentaries wherein there is no audio, only the words are coming, history channel. This is text-based search engine. Yeah. Right? So it is the commentary-based search or audio-based search rather than saying the video-based. Yes, actually when we are searching, we will, the search engine will search by the subtitles only. This is sure, because whenever we are searching the file just we are given only text-based. Now how? It is search engine. There is no middle-media. Subtitle. Okay. There is no middle-media. So one more question. When you are using the word subtitle, subtitle, I have got one more question. Whether the subtitling has been done at the time of editing or it is converting audio to text? Whenever it is uploaded only subtitle will be converted and when you load this file automatically. My question is separate. What I am asking that there is a provision in the video editing that whatever the person is speaking, the similar subtitles are created and put within the video also on a separate timeline. Right? Yes. So whether it requires that kind of video or this is a normal video. Normal. Normal. Normal any video if you upload the software will generate its particular subtitle file and that particular file will be shown. Based on audio. Yes, based on the audio. It will convert the video into audio then audio into text and then it will show the subtitle. Okay. There is nothing automatic about it. Okay. So I have got a set of about 40-50 videos. I think about 10-12 courses. Okay. With hundreds of videos. Okay. The job is to get them into a searchable format. They are going to use the software. You look at what they are claiming. They should say. They are not saying. Now imagine if four students who are working in a group. That is not what Dr. Phataka said. Why don't you say that? There is certain accuracy of what software you are doing. That is what my next question is. Correct. What they should have done. I don't know whether they have done it or not. Is a way by which a human being actually creates that. There are a lot of people who are in IIT Bombay X currently engaged in looking at this video and typing it manually. It's a waste of time. I cannot do it for past data. So what we have done is we have given them a tool which gives 70% 80% I don't know what accuracy. I don't care what the accuracy is. The software is not responsible for the audio. Who is responsible for the audio is the content management team. They have to sit here. They have to listen to Dr. Phataka and type it in. I am typing it for them. That FOR should change to FOUR manually. I am not doing it. So where is the interface for which I can do it? This is the interface only. This is the interface. You can simply go and. Go and change four to four. Suppose if you change this entirely. Why are you changing that? I am just giving an example. Don't give an example of error. Okay. Then what should I change? Go and look at it and change. You have changed or not? Why are you showing a video which you have not changed? Have they changed or not? No. They were supposed to give a complete course. That's why there were three of them and three courses. Two people should have looked at Dr. Phataka's video and changed the text to shoot the video. That was part of their responsibility. They are supposed to clean the data they have generated. They cannot walk away from the internship by saying that we have generated this okay. So that whatever two to five lines is too small. Okay. We will get thrown out in a job if we actually deliver only five lines. Okay. That is wrong English. They have to correct it. If they have not used the software which they have done. The reason why I said they have to do it is I don't want people to give unusable software. If the interface is bad, after they have done the donkey work of editing it, they will know whether the interface is good or bad. If you have not done it, then they have not completed it, get it done. Okay. From them. They can go home, do it and send it here. The internship is not complete until I get two sanitized courses from two of you. Okay. If you want to stop the certificate being given to that Amazon fellow who ran away in four weeks, ask him to at least do the donkey job before he joins Amazon. If you want. Okay. But that is not necessary because we are not giving the completion certificate. So let's not trouble him. These two are getting a certificate. You have to do sanitization of data. Not few videos. Full course. Okay. Okay. The output expected from each of you was one searchable course. That means looking at these more and more expensive languages. I don't know whether it luggage languages, what. Okay. Budget for doing the same things you write, you to write less code. This is not what he has said. No, I don't care what he has said. Have you checked it? No sir, we haven't checked it. Then you are not completed. Okay. Okay sir. Whatever interface has been given for that thing, unless you use your own software, you would have delivered some junk. I don't want the content team coming and telling me what idiot thing they have done. I don't want to rewrite your software. No sir, we have shown this to content team. You may have shown, has anybody taken one course and edited it? Till that thing is not done, I don't accept it. The proof of the pudding is in eating. Okay. You haven't answered him question that somebody has to supply keywords. Why do you think you are doing computer programming and not thermodynamics? We purposely insisted that since you are computer science people, you will begin computer science courses. Okay. Because you can edit it. Donkey work was part of it. You are supposed to edit it. Because you are computer science. When thermodynamics course comes, they are going to ask the T.S. to edit it. The T.S. will come back and say the same thing that you are doing. Okay. And I will not host the course on the searchable till the professor who gave the thermodynamics course appoints somebody to clean it up. It's not my job to clean it up. No, I can say that you have to clean it up. Only when I am convinced that there is a, there is interface is good enough. I should be able to make the assessment. Nothing. My intern had done it. Why are you talking nonsense? Okay. There is so much thing. If finally our output is all the old data has to come searchable. You have not taken the step that is good enough for me to proceed. So I have nothing to go and tell the thermodynamics fellow or the signals processing fellow that is your job. Although I am not going to do it. I have gone through this before. Doctor, I don't know you should clean it up. I will not clean up. Okay. I will not clean up somebody else's data. It is part of the responsibility of faculty. Okay. To give me a file of what he speaks. He has got TAs. He can get it done. If he doesn't get it done, his data doesn't come. But I cannot insist on that unless I am sure that it can be done. And you have not given me the guarantee. Sir. Anyway, it doesn't matter. Sir, this particular web page is going to be used by content management team. Right? This particular web page is designed for them. You are wrong. It is not going to be used by content management team. How will they use it? I am not going to spend my time content management listening to that. It is the job of the professor. Okay. Content management team will do certain work. Content management. I don't want to do some junk where content management have missed out. This is not equal to that. And they missed out the word not. This is equal to that. And that thing will go. How can content management team be responsible for SITR? Do they know thermodynamics? Do they know programming? They will only solve the English problems. That the software has not done because we have got Indian language speakers, Indian accent speakers. The software is designed for phirangis for phirangis. It doesn't work for India. So, yes. Hundred percent. Seventy percent comes. That seventy percent has to be done to hundred percent before I display it. I was told that you have done, you have not done. Secondly, you have not answered him a question. Who is going to supply the keywords for search? So, we are going to answer that. He asked you so many times. You didn't answer. You didn't answer. He asked you so many times. Sir, he... Where is metadata? That is what you have to say. The professor's TA will say what my thing will be searched for. And that piece of software has to be there. Where somebody says, This is a video. This is a video of thermodynamics. These are the keywords. And that should be packaged search. So, that keyword thing is present here. Sir, he can directly go and add keywords here. See, the metadata and keywords are nothing but the same thing. So, I asked three times to you. Sir, we did not understand the keyword. Okay. Metadata is more a scientific word. He was supposed to be given by the professor. Let's put it that way. Okay. Those keywords... Since I cannot vote Dr. Fattak to get keywords. Some TA will... No, no. He is a CS. Okay. That's why CS things were given. One of the things for them to do was programming, they can go and search on the network and do that. Okay. And keywords is okay. Not an issue. Have you done it for the two courses? Yes, sir. We have added two keywords. Few keywords. Okay. Go to the... Go to the Dr. Fattak's course. So, this is his course only? His course. No. I want to search in his course. How many times his pointers is there? Or show me the keywords. All keywords. Show me what keywords are there in... If you go to the advanced search and simply write pointers, it will show all the proper searches. This is the advanced search. What is all the proper searches and non-proper searches? The keyword search is basically, if you have like put the keywords for that particular file, assign those keywords for that file, then only it will show in the keyword search. This is the advanced search. It will show everything. From based on text? Yes. From based on text. So, even if... What way is this ordered? Result 1 to 10 of 11. In what way result 1 is differentiated from result 2? Sir. By the uproading thing, it means... It's final uproading first. No, sir. It's ridiculous. Sir, indexing that ranking thing we have not done. What thing? Sir, you told you write in the middle of the internship. Correct. Sir, but for implementing that, like you... I told you that it requires page rank algorithm and hierarchical data structure. Hierarchical data structure we have not implemented. That cannot be done, yes. When I talk functionality, I don't want to listen to keywords. Okay. It will come by more number of words. So, what you are saying is, we are displaying back when it started first. Not acceptable. It was never acceptable to me. Like number of words in the file, sir. What you use? You use database, you use table, you use hierarchical. I don't care. There is a user requirement. And that I will not go away from. Sir, it... The user requirement is when I do a search, I should get a search which is needed by the user. Yeah. In this case, it will show all the results for pointer first. After the pointer, it will show for pointers. Even pointers should be ordered. If all the results are 100, then... In one file, I say pointer. Sometimes I say point only. That also is taken because you use whoosh or push. So, point won't be shown here. Whoosh is here. Whoosh doesn't show point as pointer. No, sir. What is whoosh? Sir, it won't show for the other words. If you are searching for pointers, it will show for pointers and other words. Related to pointers. Point is not related to pointers, right? But even if pointer is spoken 100 times in 100 videos, I want to know the video which talks about pointers. That should come first. Which talks more first? It will come first. That's what I was. What is the difference between 1 and 2? Is it frequency of occurrence? Yeah, frequency of occurrence. He said it first type. No, if it is present in the title, it will show the first. I don't understand why it is shown. Why indexing and ashing is shown 2 types. Go back. Indexing and ashing. That is topic name, sir. That is topic name. The workshop team has asked us to... The advance search, no? Yeah, they have asked us. This is topic name. There is no file which has name as pointers. I don't care about file. I want you to go through all the SRT and subtitle as you call it. And find out where is pointer. Yeah, it is pointing most of the time. Why is the same thing or is it different? One is mp4, another is pdf. There are two mp4 there. This is just a topic name. This is a different file. No, no. Name of the file is different. That's the third one. The topic name we are showing here. Third one. Second and third you say. This is different. This is different. The lecture said... This is A, this is B. In this B lecture, there are pointers. How do I know? What? That one name is different, no? You are claiming. Yes, sir. Okay. That in file number 2. Forget that. In pdf file. Okay. Pointer has occurred more time than in the number 2 file. Yes. How do I know? Sir, because we have implemented so... How come? You just prove it. You take another pdf and show. Sir, what should I show? The number of counts of pointers. If it is a higher... Sir, we have seen it. We will ask them to hide. Previously, when the search was there, the number of frequencies... The frequency was shown here. But it is not the way we use search engine. Right? So, we removed that. Sir, you only ask. We can show the word count. What? Google shows the content. Yes, Google shows the headline of the file. Then, the URL. And then, some information, basic information. Where it is present and how it is used. Right? Which algorithm you are using? Lattice search? No. We are using just api-whoosh. This is online api or it is a library? Python api. Python api. So, the algorithm you don't know. What exactly happening in the behind? Fine. I think that we are running out of time. The other guys has to go for lunch.