Intro
[Intro video] The Hyperaudio Lite Editor makes audio and video more accessible by creating captions and interactive transcripts. The third, choose from several speech-to-text engines. On device, there's a Parakeet and Whisper model.[Mark] This is the Hyperaudio Lite Editor. It allows you to transcribe audio and video on your machine or from a service that will transcribe for you.
Transcribing your audiovisual media
You can do this locally, as I said, with Parakeet from NVIDIA or with Whisper. And then we have all the different Whisper models you can choose from and some language choices. Or in the cloud, you can use AssemblyAI, Deepgram, or Parakeet. On Hugging Face. You'll need keys for these, which you can get going to visiting their website. So let's choose a file.Here's one I prepared earlier. While that transcribes I'll show you a few of the key features.
Timecodes, Hide Video, Settings, Redaction and Gap Removal
We can turn timecodes on, we can turn the video or the audio off in this case. We also have a settings page, all sorts of settings here, and we can Navigate just by clicking. Turn the volume up. Turn, strike out filler words or any unwanted text and remove gaps and silences. So the 10-minute file I just submitted has been transcribed on my machine, and now I'm going to show you how to edit it.
File Management
So first let me show you a few of the file management features.Hovering over will just give you a quick view. This one here underneath is just audio, so that's why it shows like that. So you can also— you can star things. If you want to group them, unstar. You can rename, duplicate, and delete. You turn on timecodes.
Editing your Transcript
We have speakers here. I could add a speaker here in square brackets. That gets picked up. Turn speakers off now, you'll see that disappears.The main thing is to, to edit the actual text without losing the timings. We have a number of features. Let's just play it.
[Teon] Oh wait, auto advance. Okay, that was even before I was ready. Okay, cool. I'm gonna go back a slide. Hi everyone, my name—
Redaction
[Mark] Okay, as you can see, the word timings look pretty accurate, which is super useful, but we don't need this first bit. So let's play this from the start.[Teon] Hi everyone, my name is Teon Brooks.
Find and Replace
[Mark] Also, one of the main things you might find yourself doing is removing any sort of filler words. So you could do that with the find and replace tool. If I search for space, you space— oh, H space. I can replace all of those, and that's that done.On the right, we can copy the whole transcript just as text. On the left, we have a little bit of project info.
So I could just go through and correct. Here, for example, Teon is spelled Tian.
That would be one part of the correction there. Generally, the transcription is good using ParaKeet especially.
It's fine to just delete whole words. Once you're pretty happy with it, you can do a number of things. You might just want to create an interactive transcript.
Captions and Subtitles
You might just want to download the captions. So there's kind of 2 views to this. This is the transcript view, and here is the captions view. And here you see everything kind of split into subtitles. We have a measurement here that's characters per second.In your settings, you can change some of these things, so Maybe I prefer words per minute. So here I've got it so it's flagging any captions over 17 characters per second or over 180.
You can establish these figures from your own organization or your own preferences, but it just gives you a little warning by coming up in red and bolded that these might be going a little bit fast. There's a few things you can do. You can change the start and stop time. I'll just play this part here. This is a talk that I just gave at a symposium that I was So maybe I'd want to extend that time a little bit, and if I did do that, I could maybe get it down.
To a satisfactory level. So this jumps straight to the cue, this plays a clip as we've seen, we can merge, we can insert, and we can delete.
So that's kind of the captions. You see these represented here. This looks long now because I've merged 2 together.
Gaps Removal (non-spoken audio)
Another thing you can do— oops— is remove any gaps.So I've managed to shorten this by about 10% actually because About 10 minutes, and we can get rid of a minute. So significant, and I can tweak. So I'm basically removing the gaps that are longer than 500 milliseconds, but I can tweak that if I want, be more aggressive, have a slightly smaller buffer either side. Tempting fate, really.
And now if I play this back, this will be slightly quicker and there will be the occasional jerkiness in a good way. Let me just— I haven't shown you this feature.
Picture-in-Picture
This is the picture-in-picture. Can be handy. Let's play.
Live Playback of Edits
[Teon] Hi everyone, my name is Teon Brooks. This is a talk that I just gave at a symposium that I was leading last week. I was trying to do some edits that did not happen, so I'll be going through this again and updating it for this conference. So as I said, my name is Teon Brooks. I'm the president of the Gotham Data Clinic.[Mark] So you might have noticed that it kind of jumped a little bit. That's because it was removing gaps.
Saving and Export
You can save this at any point, by the way. Okay, so that's done now.I've downloaded the zip and opened that in its own folder, and what we can see is that we have the HTML, which is interactive transcript HTML. We also have the video itself and the VTT file. So let's have a look at the video first.
Let me just size this appropriately and put it up on the screen. So let's have a look.
Reencoding from within the Browser
[Teon] Hi everyone, my name is Tian Brooks. This is a talk that I just gave at a symposium that I was leading last week. I was trying to do some edits that did not happen, so I'll be going through this again and updating it for this conference. So as I said, my name is Tian Brooks. I'm the president of the Gotham Data Clinic, and we are a new nonprofit based in New York City. We just received our 501, so we are sprinting ahead. So I'm just taking the show on the road, and I'll tell you about some of the work that we're doing.So that's one of many—
[Mark] So there you go, you can see that that's quite easy to understand at that speed. Some really nice captions burnt on with sort of karaoke effect, great for social media. And it's removed any sort of gaps, and you can see that, and also any sort of that's been done. So that's the video as a 7 and 8— an 8-minute file, 7 and a half minute file.
Interactive Transcripts
And we've also got an interactive transcript which can play.[Teon] Hi everyone, my name is Teon Brooks. This is a talk that I just gave at a symposium that I was leading last week.
[Mark] I'll try and do some— And also the captions, which we don't have to look at right now. So that's pretty much everything. You might want to explore all the different type of files you can actually export. You can have karaoke-style WebVTT if you want to and not have it burnt on at all. That will only work in Chrome and Safari, that won't work in Firefox, by the way. You can also save it as a HyperAudio file, which is portable. You've got your classic TXT, Markdown, DOCX, JSON, Hyperaudio JSON. This Ionosphere AT protocol is a new decentralized protocol.
AT Protocol
You can actually publish this to a PDS, which is kind of cool. You need to know, you need to give it the URL of the media, so I haven't got that just now because I've got it local on my machine, so But you basically, within something like Bluesky, can just create a new password, put your own handle in here, and you can publish on it on PDS and just share that link with anyone you want to and unpublish it when you're finished. I don't think there's much else. We can toggle the sidebar. Like I say, we have sort of picture-in-picture happening.When we need it, and that can even go on a separate screen. We've got find and replace, which you've seen. So yeah, that's, that's it basically. That's a whistle-stop tour of the Hyperaudio Lite Editor. Give it a whirl, and I'd be excited to hear any sort of feedback or ideas that you might have.