How to Use Whisper for Translating Audio to English Subtitles in Subtitle Edit
Learn how to use Whisper for translating audio into English subtitles in Subtitle Edit. Discover its performance across languages and practical tips.
File
How to Use Whisper to Automatically Translate Audio or Video to English Subtitles - Subtitle Edit
Added on 09/05/2024
Speakers
add Add new speaker

Speaker 1: In this video, I'll show you how to use Whisper to automatically translate your audio from one language into English subtitles in subtitle edit. Welcome to this video. My name is David. This is still experimental in subtitle edit. And it's still a feature that is still growing every day outside there in the world. Now I'll take you to the documentation first. And I'll just let you know here that it's a robust speech recognition via large weak supervision. And the reason why we may need to use this is you may need to translate audio from like Spanish into English subtitles. Now, the only issue we have, if I scroll down slightly, is that Whisper's performance varies widely depending on the language. The figure below shows a wide error rate breakdown by languages of flused data set using the large model. More word error rate and blue scores corresponding to the other models and data set can be found in Appendix D in the paper. Now, you'll notice that some of the languages that have very low word error rate like Spanish, Italian, English, Portuguese, German, Japanese, Russian, Polish, French, Catalan etc. Like under like 10%. This is probably what will work best for you. For my audio that I have, it's Swahili and we have a word error rate of about 47.9%. Now, that's a huge error rate. You can see it here. Swahili somewhere here. 47.9. So I can expect that the translation is going to be garbage. So this is how you do it. This is the beta version. So it's not in the 3.6.8. So if I check about, you'll see that it's subtitled at 3.6.8 next comma beta 51. So we can do this go to video, open video file. And then I'll select the Swahili audio. But I'll not take you through the agonizing process of having to wait for Whisper to automatically transcribe this Swahili audio and translate it into English. I have another instance that I've done that. Click on open. Now once you do that, go to video, audio to text by Whisper. And then from this section, select the language of this particular audio. S to go down and then we can select Swahili. Now once you do that, the translate to English option becomes active. Click on that. That's awesome. And then click on generate. I'll not take you through that agonizing process because it takes a long time to transcribe the Swahili audio and then translate it. Now I'll just take you to an instance that's already done. And I can honestly tell you, this is not what I say in that particular audio. So if I actually play it.

Speaker 2: So, how do you protect yourself from swine fever? Many people protect themselves by either looking for a job or by continuing with their lives. However, there are people in the field who are in a difficult position to get a job. Especially young people who have found a way to build themselves with the knowledge they have. Many farmers have a habit of plotting their crops. If this habit of plotting your crops continues, you will be more likely to be killed.

Speaker 1: Now simply put, whatever is written here is not what I've talked about. Because I'm talking about the way people are shifting from the rural areas to the urban setting. More or less how farming land is becoming scarce. And then there's something like more even of students who are studying in the universities and something of that sort. But the transcript I got here, which is actually limited to just a particular section is not what I mentioned. So for languages with word error rates that probably above 10% ish, maybe you just have to try it out, you'll have a very poor translation from that particular language into English. But I've seen people who are doing Japanese and it's almost perfect in terms of translation. So you can go test it out. But that is how you can use the audio to text feature in Whisper to translate the language that you have into English subtitles. Now that's it from me. I hope this video is of value to you. Thank you so much for watching.

ai AI Insights
Summary

Generate a brief summary highlighting the main points of the transcript.

Generate
Title

Generate a concise and relevant title for the transcript based on the main themes and content discussed.

Generate
Keywords

Identify and highlight the key words or phrases most relevant to the content of the transcript.

Generate
Enter your query
Sentiments

Analyze the emotional tone of the transcript to determine whether the sentiment is positive, negative, or neutral.

Generate
Quizzes

Create interactive quizzes based on the content of the transcript to test comprehension or engage users.

Generate
{{ secondsToHumanTime(time) }}
Back
Forward
{{ Math.round(speed * 100) / 100 }}x
{{ secondsToHumanTime(duration) }}
close
New speaker
Add speaker
close
Edit speaker
Save changes
close
Share Transcript