Speaker 1: Hi, everyone. In this video, we'll talk about AWS Transcribe. In this video, we'll go over an introduction to the Transcribe service from AWS, including general facts and usage. Then we'll move on to a demo, including custom vocabulary, vocabulary filters, and Transcribe jobs. AWS is an ecosystem of cloud computing technologies by Amazon, and Transcribe is one of them that integrates very well with the AWS ecosystem. It can do many things from transcribing customer calls to captioning and subtitling. What is AWS Transcribe? Amazon Transcribe is an automatic speech recognition service that makes it easy for developers and customers to add speech-to-text capability to applications. It's great if you want to convert audio into text and don't have time to write it yourself or don't want to hire a transcriber. It gives you automatic speech detection using machine learning, which includes live voice typing. Since AWS Transcribe uses machine learning to do their transcription, it's different from other transcription websites and services because it'll have correct grammatical formatting and punctuation. This puts it high on the list of transcription services. AWS Transcribe first came out with just English and Spanish, but lately, they've added many, many languages, ranging from Spanish to German. This makes it even easier to do subtitling and captioning. It also integrates into many technologies, such as Python, Java, .NET, Go, JavaScript, PHP, Ruby, and many more. It also integrates with many AWS technologies for real-time transcription. I wanted to reiterate how many languages are available for use in AWS Transcribe, so here they are. Let's talk about its usage. Some of the uses of AWS are transcribing customer calls, if you can't be there to talk to a customer, then you can use AWS's service to transcribe the call and take a look later. It's also great for automating closed-caption subtitling if you don't want to write it out yourself. AWS Transcribe Medical can be used for transcribing clinical documentation applications, and we'll take a look at that in the demo. All right, let's get to the demo. So in my last video on S3, we went to aws.amazon.com, and that's where we're going today also. So I'm going to sign in to my console first. All right, here it is. And there are four ways to get to AWS Transcribe. If it's in your recently visited services, you can go there, or it'll be in the machine learning section. Here it is. Or you can search it up, transcribe, and it filters it up for you right here. And the last way is that you can bookmark it. So here are all the technologies made by AWS, and you can find Transcribe here. You can just drag it over, and you can drag over whatever else technologies you want, and then click this to go out. Now let me go to Transcribe. First, we're going to talk about real-time transcription. It listens to your real-time and also gives you the text as you speak. So let's try it out. You can choose your language of choice, but I'll keep it as default. So now if we press Start Streaming, then we can start talking and it'll transcribe it. Amazon Transcribe is a great way to transcribe your audio with ease. And if I press Stop Streaming, then as you can see, it worked. Cool. Now let's go on to Custom Vocabulary. Custom Vocabulary, let's leave that. Custom Vocabulary lets you add words or names that may not be known to AWS Transcribe. That's for names or words that are not commonly used. I've already made one for my name, and we can test it out by clicking this. And now if we press Test in Real-Time, then it goes to the real-time transcription. And in Additional Settings, you can see that we added Custom Vocabulary. So if I press Start Streaming and say, Hi, my name is Rishab. Then it should say it. Yes, as you can see here, because I put the Custom Vocabulary of Rishab, it worked. Now if we take it out, then let's try it again. Let me reload the page so that it takes out the... Okay. Is this gone? Yes. Hi, my name is Rishab. Now, see, it doesn't work. That proves that Custom Vocabulary is a success. Now let's move on to Vocabulary Filtering. Let's leave this. Vocabulary Filtering filters out words you don't want and doesn't include them in your text. This feature helps us eliminate profane or unwanted words from our transcription jobs. Now let's make a transcription job. So I'll go to Transcription Jobs. And as you can see, I already have two. But for this one, we're going to create a new one. I'll call it Rishab Teaches Tech. All right. And it's asking us for language. Again, there's plenty of them, but I'm going to keep it as English. Now it's asking us for an audio in the form of an S3 file. So I already have an audio file in an S3 bucket. So let me do that. It's this one. And we can click this and press Choose. The files must be in one of these formats. WAV format, MP3 or MP4, or a FLAC file. And they all have to be stored in an S3 bucket. So next is the output data. After the transcription's done, output data is the location where Transcribe stores the text. The default option is the Service Managed S3 bucket, and the output will be removed after 90 days. But if you choose Customer Specified, it'll save in your S3 bucket. As you can see, it's asking me for my S3 bucket, but I don't need that. It'll also ask us for encryption, but I'm not going to get into that now. So let's go back to Service Managed and press Next. Now you can configure your job, but I'm not going to do anything. I'm going to keep it the same. And press Create. As you can see, it says Transcription job successfully created. But it's saying the status is in progress. For it to become complete, it'll take a bit of time. It took about two minutes to complete, but as you can see, now it's completed. So if we press it, then we can see the input data location and all the job details. Now if we go down, then the transcription preview will be here. Here is the text. This recording... Oh, as you can see here, if you hover over each word, then it tells you the confidence. So yeah. It says this recording is for the Amazon Transcribe video by Russia. Yeah, so everything was correct except Russia. I said Rishabh instead of Russia. If we want, we can put custom vocabulary on this one. And if we hold, it just said Russia instead. This application integration instructions provide us the details of how to initiate Transcribe audio and video in real time with your application. The same Transcribe job we just created manually could be automated by calling an API. And here are the instructions. This is the API payload in which I am mentioning the media format and the file URI. If you call this API programmatically, response JSON will be in this format. So if we copy this URI here, it's a very long URL. So if we copy it all the way down here and open it in a new tab, then it downloads a JSON file. So as you can see here, this is all the contents that was in our Transcribe job. As you can see, this is the transcript. This is what we got before. And yeah. There's also an SDK available for Python and a few other languages to automate Transcribe job creation and management. So let me check it out. I'll put this link in the description. And there's for Python and a few other languages. Let's move on to AWS Transcribe Medical. Let me go here. And Amazon Transcribe. Launch it. It's a service that makes it easy to quickly create accurate transcriptions from medical consultations between patients and physicians. The difference is that it includes medical and pharmaceutical terms used in clinical documentation, all in dictated notes. There's also the three same services that are in the regular Amazon Transcribe. So as you can see, there's conversation dictation. There's a little more options for AWS Transcribe Medical. But yeah, that's pretty much it. This concludes the AWS Transcribe demo. We looked at an AWS service and its features with a quick demo. Thanks very much for watching. If you all had any doubts, please comment down below. I would love to help you out if you're stuck with any AWS Transcribe questions or issues. Please like, subscribe, all that jazz. Until then, you can learn anything.