Creating a Multilingual API with Bisper Model (Full Transcript)

Speaker 1: Hello friends, in this video we will discuss how to make a transcribe api In this we are going to use the bisper model The medium model of the bisper model The multilingual models we will use in this So basically we have to install the bisper model So it has some packages Requirements will be ff, mpeg, python In the setup we will install the bisper model Bisper library And after that The supporting library We will install it then after this we are using flask in this to make api we are loading the bisper model here and after this we are importing name temporarily file from temp file because of which if any file comes to us then we can dump that file temporarily when our processing is complete after that it is automatically deleted from any memory right, so what we have done in this we have created an object of flask so we have kept the starting route point as a slash and the method we have used is POST so basically we will upload the audio files to the API so we are going to use the python request request.files so if there are no files in this then it will return about 400 so basically you have not passed any file in this so it is a bad request we will process these files one by one and we will make a dictionary and append it at the end, we will dump our result in JSON and we will send the response back after that we are calling the last miss so basically when this file will run then our if statement will start running so we have defined the port number 5001 debug equal to true if any code changes then it will automatically reload the file ok so let's run it and pass an audio file in it that it detects English or not or detects any Hindi language and converts it into Hindi or not so we will type python app.py what will happen with this server will be up then after this we will use postman collection we will use postman through which we will send request on this api ok so in this we will go to bodies in bodies we will pass the file ok so file 1 we are passing .web ok it can be in any format it can be mp3 also it can be .web also right so our server is up right so now what we will do on this we will send request on server so as you can see we are selecting file ok multiple files we have sent because we are processing it in for loop all the files you upload will be processed one by one and append it in the dictionary and JSON result will send it to us ok, now we will send so now the process is going on in the background see I have written my name in Hindi Yoginder Singh translated right now we take any other file ok now we will process it now you can see how quickly it has returned right you can adjust your microphone settings using standard it has done a little wrong here it should have come here add of player tools right decreasing echo and adjusting the volumes after that I said thank you it has translated correctly so in this way you can make any transcribe api and you can use it anywhere in your product or you want to make a service somewhere then you can fine tune this bisper model on your data or you can make it in product and give it to anyone right so how did you like this video do like, subscribe and share and stay connected with us stay safe, stay tuned, thank you guys