Beauty Box 6.5 introduces new facial spline capabilities that turn AI face-tracking data into editable After Effects masks. Let’s take a look at what makes facial splines different and how you can use them in your After Effects workflow.
Adobe Premiere Pro’s AI Assistant + Beauty Box: Prompts to Try
Adobe Premiere Pro’s new AI Assistant is now in the BETA version of the software! It promises to not only organize your footage, but also to speed up editing but how well so… We got curious and decided to test if it works with third-party plugins!
Photoshop’s Generative Fill Isn’t Great, But It Works Well at Fixing Other GenAI Images
One problem with generative AI is that it’s difficult to get exactly what you want. You can often get something that’s good enough but more often than not, you get 90% of the way to what you want and getting the AI to make the correct changes to get to a 100% is daunting. (which is why I’m a little skeptical about GenAI for video. For generic B-roll stuff, sure maybe, but wrangling the correct prompts for a 30 second video that needs to be exactly this or that is going to be difficult to say the least. It’s hard enough for a single still image.) Photoshop’s Generative AI (called Generative Fill) is pretty subpar when compared to some of the more cutting edge ones (DALL-E, Stability AI, etc) for creating images from scratch. However, what it does pretty well is extending images. i.e. If you’ve got an image that you want wider or more head room than it was originally shot with. OR… if you’ve created something with another AI tool, like DALL-E, as I’ve done here. DALL-E gave me more or less what I wanted but without much of a tail. I spent another 20 minutes or so trying to get DALL-E to give me this fish with a tail before giving up. It really wanted to redo the entire image. So it got frustrating. This is where Photoshop’s GenAI turned out to be useful. To be fair, they market it as more of a way to extend/improve existing images than creating stuff from scratch. It can create from scratch but the results often aren’t great. But when it comes to extending images, there’s a big advantage to being in Photoshop… selections! You can make the canvas wider, select the empty area to the side and type in ‘extend image’. Boom. Now of course it gave me some other variations that didn’t work at all, but doesn’t matter. It gave me a great variation that did work. Also, prompting something like ‘extend image with skeleton of an angler fish’ didn’t work. It was the simpler prompt ‘extend image’ that did the trick. (Prompts are weird and a whole art unto themselves. Figuring out what the AI is going to respond to takes a LOT of trial and error. And then you still need to get it to do what you want.) I then selected the other side and it created that easily. You can see slight seams where the image was extend. When having Photoshop create the extensions, I tried both selecting the area by itself and selecting a little of the original image (including feathering it). It didn’t really make much difference. You got slightly different imagery but the seams tended to show up no matter what. The tail was the worst problem however. There was an obvious change in style from the original to the Photoshop extension. So I selected just that bit and ran Content Aware Fill a few times to cover up the seam. And that worked reasonably well despite CA Fill not being AI. It’s just sampling from other parts of the image. Selecting the seam and running Generative Fill (prompt: ‘remove seam’) on it created three variations. Two of the three didn’t work but the third one arguably looks better than CA Fill. But they’re both pretty good. So just realize CA Fill can help touch up slight imperfections as well. Getting DALLE, Midjourney or whatever to give you exactly what you want can be difficult. If you get most of the way there, but are having trouble prompting those tools to fill in the details, Photoshop’s Generative Fill may be able to touch things up or extend the image more easily. Here’s the final image:
Free NAB 2024 Exhibits Pass, Come See Beauty Box 6.0
NAB is three months away! We have a booth right next to Adobe in South Hall Lower, SL7043, so we’ll be easy to find. If you haven’t registered yet, you can use this code for a free exhibits pass: NS9892 Go here to register for NAB: https://registration.experientevent.com/ShowNAB241/ We’ve got a new version of Beauty Box, our AI based digital makeup and retouching plugin, coming soon! It’ll be released before NAB, but we’ll definitely be showing that off. The big new feature uses AI to whiten teeth. We’ve always used AI in other ways in Beauty Box, but this is a much more sophisticated version. We’ve put a lot of work into it and we’re pretty excited about it. There will also be a couple other announcements/release around April, so stayed tune for that. If you haven’t chatted with us lately, come by the booth and say hello. You can learn about all the new stuff. Including Data Storyteller which we released a few months ago and won a Best of NAB 2023 from Videomaker!
Testing A.I. Transcript Accuracy (most recent test)
Periodically we do tests of various AI services to see if we should be using something on the backend of Transcriptive-A.I. We’re more interested in having the most accurate A.I. than we are with sticking with a particular service (or trying to develop our own). The different services have different costs, which is why Transcriptive Premium costs a bit more. Gives us more flexibility in deciding which service to use. This latest test will give you a good sense of how the different services compare, particularly in relation to Adobe’s transcription AI that’s built into Premiere. The Tests Short Analysis (i.e. TL;DR): For well recorded audio, all the A.I. services are excellent. There isn’t a lot of difference between the best and worst A.I… maybe one or two words per hundred words. There is a BIG drop off as audio quality gets worse and you can really see this with Adobe’s service and the regular Transcriptive-A.I. service. A 2% difference in accuracy is not a big deal. As you start getting up around 6-7% and higher, the additional time it takes to fix errors in the transcript starts to become really significant. Every additonal 1% in accuracy means 3.5 minutes less of clean up time (for a 30 minute clip). So small improvements in accuracy can make a big difference if you (or your Assistant Editor) needs to clean up a long transcript. So when you see an 8% difference between Adobe and Transcriptive Premium, realize it’s going to take you about 25-30 minutes longer to clean up a 30 minute Adobe transcript. Takeaway: For high quality audio, you can use any of the services… Adobe’s free service or the .04/min TS-AI service. For audio of medium to poor quality, you’ll save yourself a lot of time by using Transcriptive-Premium. (Getting Adobe transcripts into Transcriptive requires a couple hoops to jump through, Adobe didn’t make it as easy as they could’ve, but it’s not hard. Here’s how to import Adobe transcripts into Transcriptive) (For more info on how we test, see this blog post on testing AI accuracy) Long Analysis When we do these tests, we look at two graphs: How each A.I. performed for specific clips The accuracy curve for each A.I. which shows how it did from its Best result to Worst result. The important thing to realize when looking at the Accuracy Curves (#2 above) is that the corresponding points on each curve are usually different clips. The best clip for one A.I. may not have been the best clip for a different A.I. I find this Overall Accuracy Curve (OAC) to be more informative than the ‘clip-by-clip’ graph. A given A.I. may do particularly well or poorly on a single clip, but the OAC smooths the variation out and you get a better representation of overall performance. Take a look at the charts for this test (the audio files used are available at the bottom of this post): All of the A.I. services will fall off a cliff, accuracy-wise, as the audio quality degrades. Any result lower than about 90% accuracy is probably going to be better done by a human. Certainly anything below 80%. At 80% it will very likely take more time to clean up the transcript than to just do it manually from scratch. The two things I look for in the curve is where does it break below 95% and where does it break below 90%. And, of course, how that compares to the other curves. The longer the curve stays above those percentages, the more audio degradation a given A.I. can deal with. You’re probably thinking, well, that’s just six clips! True, but if you choose six clips with a good range of quality, from great to poor, then the curve will be roughly the same even if you had more clips. Here’s the full test with about 30 clips: While the curves look a little different (the regular TS A.I. looks better in this graph), mostly it follows the pattern of the six clip OAC. And the ‘cliffs’ become more apparent… Where a given level of audio causes AI performance to drop to a lower tier. Most of the AIs will stay at a certain accuracy for a while, then drop down, hold there for a bit, drop down again, etc. until the audio degrades so much that the AI basically fails. Here are the actual test results: TS A.I. Adobe Speechmatics TS Premium Interview 97.2% 97.2% 97.8% 100.0% Art 97.6% 97.2% 99.5% 97.6% NYU 91.1% 88.6% 95.1% 97.6% LSD 92.3% 96.9% 98.0% 97.4% Jung 89.1% 93.9% 96.1% 96.1% Zoom 85.5% 80.7% 89.8% 92.8% Remember: Every additonal 1% in accuracy means 3.5 minutes less of clean up time (for a 30 minute clip). So that’s the basics of testing different A.I.s! Here are the clips we used for the smaller test to give you an idea of what’s meant by ‘High Quality’ or ‘Poor Quality’. The more jargon, background noise, accents, soft speaking, etc there is in a clip, the harder it’ll be for the A.I. to produce good results. And you can hear that below. You’ll notice that all the clips are 1 to 1.5 minutes long. We’ve found that as long as the clip is representative of the whole clip it’s taken from, you don’t get any additional info from the whole clip. An hour long clip will product similar results to one minute, as long as that one minute has the same speakers, jargon, background noise, etc. Any questions or feedback, please leave a note in the Comments section! (or email us at cs@digitalanarchy.com)
Transcription Accuracy: Adobe Sensei vs Transcriptive A.I.
Speechmatics, one of the A.I. engines we support, recently released a new speech model which promised much higher accuracy. Transcriptive Rough Cutter now supports that if you choose the Speechmatics option. Also with Premiere able to generate transcripts with Adobe Sensei, we get a lot of questions about how it compares to Transcriptive Rough Cutter. So we figured it was a good time to do a test of the various A.I. speech engines! (Actually we do this pretty regularly, but only occasionally post the results when we feel there’s something newsworthy about them) You can read about the A.I. testing methodology in this post if you’re interested or want to run your own tests. But, in short, Word Error Rate is what we pay most attention to. It’s simply: NumberOfWordsMissed / NumberOfWordsInTranscript where NumberOfWordsMissed = the number of words in the corrected transcript that the A.I. failed to recognize. If instead of the word ‘Everything’ the A.I. produced ‘Even ifrits sing’, it still missed just one word. In the reverse situation, it would count as three missed words. We also track punctuation errors, but those can be somewhat subjective, so we put less weight on that. What’s the big deal between 88% and 93% Accuracy? Every 1% of additional accuracy means roughly 15% less incorrect words. A 30 minute video has, give or take, about 3000 words. So with Speechmatics you’d expect to have, on average, 210 missed words (7% error rate) and with Adobe Sensei you’d have 360 missed words (12% error rate). Every 10 words adds about 1:15 to the clean up time. So it’ll take about 18 minutes more to clean up that 30 minute transcript if you’re using Adobe Sensei. Every additonal 1% in accuracy means 3.5 minutes less of clean up time (for a 30 minute clip). So small improvements in accuracy can make a big difference if you (or your Assistant Editor) needs to clean up a long transcript. Of course, the above are averages. If you have a really bad recording with lots of words that are difficult to make out, it’ll take longer to clean up than a clip with great audio and you’re just fixing words that are clear to you but the A.I. got wrong. But the above numbers do give you some sense of what the accuracy value means back in the real world. The Test Results! All the A.I.s are great at handling well-recorded audio. If the talent is professionally mic’d and they speak well, you should get 95% or better accuracy. It’s when the audio quality drops off that Transcriptive and Speechmatics really shine (and why we include them in Transcriptive Rough Cutter). And I 100% encourage you to run your own tests with your own audio. Again, this post outlines exactly how we test and you can easily do it yourself. Speechmatics New is the clear winner, with a couple first place finishes, no last place finishes, and at 93.3% rate overall (you can find the spreadsheet with results and the audio files further down the post). One caveat… Speechmatics takes about 5x as long to process. So a 30 minute video will take about 3 minutes with Transcriptive A.I. and 15-20 minutes with Speechmatics. If you select Speechmatics in Transcriptive, you’re getting the new A.I. model. Adobe Sensei is the least accurate with two last place finishes and no first places, for a 88.3% accuracy overall. Google, which is another A.I. service we evaluate but currently don’t use, is all over the place. Overall, it’s 80.6%, but if you remove the worst and best examples, it’s a more pedestrian 90.3%. No idea why it failed so badly on the Bill clip, but it’s a trainwreck. The Bible clip is from a public domain reading of the bible, which I’m guessing was part of Google’s training corpus. You rarely see that kind of accuracy unless the A.I. was trained on it. Anyways, this inconsistency is why we don’t use it in Transcriptive. Here are the clips we used for this test: Here’s the spreadsheet of the results (SM = Speechmatics, Green means best performance, Orange means worst). Again, mostly we’re focused on the Word Accuracy. Punctuation is a secondary consideration:
How do We Test Speech-to-Text Services for Accuracy?
Transcriptive-A.I. doesn’t use a single A.I. services on the backend. We don’t have our own A.I., so like most companies that offer transcription we use one of the big companies (Google, Watson, Speechmatics, etc). We initially started off with Speechmatics as the ‘high quality’ option. And they’re still very good (as you’ll see shortly), but not always. However, since we had so many users that liked them, we still give you the option to use them if you want. However, we’ve now added Transcriptive-A.I. This uses whatever A.I. services we think is best. It might use Speechmatics, but it might also use one of a dozen other services we test. Since we encourage users to test Transcriptive-A.I. against any service out there, I’ll give you some insight on how we test the different services and choose which to use behind the scenes. Usually we take 5-10 audio clips of varying quality that are about one minute long. Some very well recorded, some really poorly recorded and some in between. The goal is to see which A.I. works best overall and which might work better is certain circumstances. When grading the results, I save out a plain text file with no timecode, speakers or whatever. I’m only concerned about word accuracy and, to a lesser degree, punctuation accuracy. Word accuracy is the most important thing. (IMO) For this purpose, Word 2010 has an awesome Compare function to see the difference between the Master transcript (human corrected) and the A.I. transcript. Newer versions of Word might be better for comparing legal documents, but Word 2010 is the best for comparing A.I. accuracy. Also, let’s talk about the rules for grading the results. You can define what an ‘error’ is however you want. But you have to be consistent about how you apply the definition. Applying them consistently matters more than the rules themselves. So here are the rules I use: 1) Every word in the Master transcript that is missed counts as one error. So ‘a reed where’ for ‘everywhere’ is just one error, but ‘everywhere’ for ‘every hair’ is two errors. 2) ah, uh, um are ignored. Some ASRs include them, some don’t. I’ll let ‘a’ go, but if an ‘uh’ should be ‘an’ it’s an error. 3) Commas are 1/2 error and full stops (period, ?) are also 1/2 error but there’s an argument for making them a full error. 4) If words are correct but the ASR tries to separate/merge them (e.g. ‘you’re’ to ‘you are’, ‘got to’ to ‘gotta’, ‘because’ to ’cause) it does not count as an error. That’s it! We then add up the errors, divide that by the number of words that are in the clip, and that’s the error rate!
Converting an SRT (or VTT) Caption File to Plain Text File for Free
This is a quick blog post showing you how to use the free Transcriptive trial version to convert any SRT caption file into a text file without timecode or line numbers (which SRTs have). You can do this on Transcriptive.com or if you have Premiere, you can use Transcriptive for Premiere Pro. This can occur because you have a caption file (SRT or VTT) but don’t have access to the original transcript. SRT files tend to look like this: 1 00:00:02,299 –> 00:00:09,100 The quick brown fox 2 00:00:09,100 –> 00:00:17,200 hit the gas pedal and And you might want normal human readable text so someone can read the dialog, without the line numbers and timecode. So this post will show you how to do that with Transcriptive for free! We are, of course, in the business of selling software. So we’d prefer you bought Transcriptive BUT if you’re just looking to convert an SRT (or any caption file) to a text file, the free trial does that well and you’re welcome to use it. (btw, we also have some free plugins for After Effects, Premiere, FCP, and Resolve HERE. We like selling stuff, but we also like making fun or useful free plugins) Getting The Free Trial License As mentioned, this works for the Premiere panel or Transcriptive.com, but I’ll be using screenshots from the panel. So if you’re using Transcriptive.com it may look a little bit different. You do need to create a Transcriptive account, which is free. When the panel first pops up, click the Trial button to start the registration process: You then need to create your account, if you don’t have one. (If you’re using Transcriptive.com, this will look different. You’ll need to manually select the ‘free’ account option.) Importing the SRT Once you register the free trial license, you’ll need to import the SRT. If you’re on Transcriptive.com, you’ll need to upload something (could be 10sec of black video, doesn’t matter what, but there has to be some media). If you’re in Premiere, you’ll need to create a Sequence first, make sure Clip Mode is Off (see below) and then you can click IMPORT. Once you click Import, you can select SRT from the dropdown. You’ll need to select the SRT file using the file browser (click the circled area below). Then click the Import button at the bottom. You can ignore all the other options in the SRT Import Window. Since you’re going to be converting this to a plain text file without timecode, none of the other stuff matters. After clicking Import, the Transcriptive panel will look something like this. The text from the SRT file along with all the timecode, speakers, etc: Exporting The Plain Text File Alright… so how do we extract just the text? Easy! Click the Export button in the lower, left corner. In the dialog that gets displayed, select Plain Text: The important thing here is to turn OFF ‘Display Timecode’ and ‘Include Speakers’. This will strip out any extra data that’s in the SRT and leave you with just the text. (After you hit the Export button) That’s it! Ok, well, since caption files tend to have lines that are 32 characters long you might have a text file that looks like this: The quick brown fox hit the gas pedal and If you want that to look normal, you’ll need to bring it into Word or something and replace the Paragraphs with a Space like this: And that will give you: The quick brown fox hit the gas pedal and And now you have human readable text from an SRT file! A few steps, but pretty easy. Obviously there are lots of other things you can do with SRTs in Transcriptive, but converting the SRT to a plain text file is one that can be done with the free trial. As mentioned, this also works with VTT files as well. So grab the free trial of Transcriptive here and you can do it yourself! You can also request an unrestricted trial by emailing cs@digitalanarchy.com. While this SRT to Plain Text functionality works fine, there are some other limitations if you’re testing out the plugins for transcripts or editing the text.
Tom Cruise and The Deepfake End of The World
Any time I hear people freaking out about A.I., for good or bad, I’m skeptical. So much that happens in the world of A.I. is hype that the technology may never live up to, much less live up to it now, that you have to take it with a grain of salt. So it goes with the great Tom Cruise deepfake. It took a good VFX artist two months to turn footage of a Tom Cruise impersonator into the videos that people are freaking out about. The Verge has a good article on it. This is an awesome demo reel for the VFX artist involved. It’s very well done. But it doesn’t make an awesome case for the technology disrupting the world as we know it. It needed raw footage of someone that looked and acted like Cruise. It then took months to clean up the results of the A.I. modifying the Cruise look-a-like. (It does make an interesting case for the tech to be used in VFX, but that’s something different) This isn’t a ‘one-click’ or even a ‘whole-bunch-of-clicks’ technology. It’s a ‘shit-ton-of-work’ technology and given they had the footage of the Cruise look-a-like you can make an argument that it could’ve been done in less time with traditional rotoscoping and compositing. Anyways, the fear and consternation has gotten ahead of itself. We’ve had the ability to put people in photos they weren’t in for a long time. We’ve figured out how to deal with it. (not to mention, it’s STILL very difficult to cut someone out of one picture and put them in another one without it being detectable… and Photoshop is, what, 30 years old now?) It’s good to consider the implications but we’re a long, long way from anyone being able to do this to any video.
Why do we have the lowest transcription costs?
We occasionally get questions from customers asking why we charge .04/min ($2.40/hr) for transcription (if you pre-pay), when some competitors charge .25/min or even .50/min. Is it lower accuracy? Are you selling our data? No and no. Ok, but why? Transcriptive and PowerSearch work best when all your media has transcripts attached to it. Our goal is to make Transcriptive as useful as possible. We hope the less you have to think about the cost of the transcripts, the more media you’ll transcribe… resulting in making Transcriptive and PowerSearch that much more powerful. The Transcriptive-AI service is equal to, or better, than what other services are using. We’re not tied to one A.I. and we’re constantly evaluating the different A.I. services. We use whatever we think is currently state-of-the-art. Since we do such a high volume we get good pricing from all the services, so it doesn’t really matter which one we use. Do we make a ton of money on transcribing? No. The services that charge .25/min (or whatever) are probably making a fair amount of money on transcribing. We’re all paying about .02/min or less. Give or take, that’s the wholesale/volume price. If you’re getting your transcripts for free… those transcripts are probably being used for training, especially if the service is keeping track of the edits you make (e.g. YouTube, Otter, etc.). Transcriptive is not sending your edits back to the A.I. service. That’s the important bit if you’re going to train the A.I. Without the corrected version, the A.I. doesn’t know what it got wrong and can’t learn from it. So, for us, it all comes down to making Transcriptive.com, the Transcriptive Premiere Pro panel, and PowerSearch as useful as possible. To do so, we want the most accurate transcripts and we want them to be as low cost as possible. We know y’all have a LOT of footage. We’d rather reduce the barriers to you transcribing all of it. So… if you’re wondering how we can justify charging .04/min for transcripts, that’s the reason. It enables all the other cool features of Transcriptive and PowerSearch. Hopefully that’s a win for everyone.
Changes between Transcriptive 2.x and 1.x
We often get asked what the differences are between Transcriptive 2.0 and 1.0. So here is the full list of new features! As always there are a lot of other bug fixes and behind the scenes changes that aren’t going to be apparent to our customers. So this is just a list of features you’ll encounter while using Transcriptive. NEW FEATURES IN TRANSCRIPTIVE 2.0 Works with clips or sequences: You no longer have to have clips in sequences to get them transcribed. Clips can be transcribed and edited just by selecting them in the Project panel. This opens up many different workflows and is something the new caption system in Premiere can’t do. Watch the tutorial on transcribing clips in Premiere A clip selected in the Project panel. Setting In/Out points in TS! Editing with Text: Clip Mode enables you to search through clips to find sound bites. You can then set IN/OUT points in the transcript and insert them into your edit. This is a powerful way of compiling rough cuts without having to scrub through footage. Watch the Tutorial on editing video using a transcript! Collaborate by Sharing/Send/receive to Transcriptive.com: Collaborate on creating a paper edit by sharing the transcript with your team and editor. Send transcripts or videos from Premiere to Transcriptive.com, letting a client, AE, or producer edit them in a web browser or add Comments or strike-through text. The transcript can then be sent back to the video editor in Premiere to continue working with it. Watch the tutorial on collaborating in Premiere using Transcriptive.com! There’s also this blog post on collaborative workflows. Now includes PowerSearch for free! Transcriptive can only search one transcript at a time. With PowerSearch, you can search every clip and sequence in your project! It’s a search engine for Premiere. Search for text and get search results like Google. Click on a result and it jumps to exactly where the dialog is in that clip or sequence. Watch the tutorials on PowerSearch, the search engine for Premiere. Search results in Premiere! Click to jump to that point in the media. Reduced cost: As low as .04/min. by prepaying minutes you can get the cost down to .04/min! Why is it so inexpensive? Is it worse than the other services that charge .25 or .50/min? No! We’re just as good or better (don’t take my word, run your own comparisons). Transcriptive only works if you’ve transcribed your footage. By keeping the cost of minutes low, hopefully we make it an easy decision to transcribe all your footage and make Transcriptive as useful as possible! Ability to add comments/notes at any point in the transcript. The new Comments feature lets you add a note to any line of dialog. Incredibly useful if you’re working with someone else and need to share information. It’s also great if you want to make notes for yourself as you’re going through footage. Strikethrough text: Allows you to strikethrough text to indicate dialog that should be removed. Of course, you can just delete it but if you’re working with someone and you want them to see what you’ve flagged for deletion OR if you’re just unsure if you want to definitely delete it, strikethrough is an excellent way of identifying that text. Glossary: Unlimited glossary for increasing the A.I. accuracy. This allows you to enter in proper names, company names, jargon and other difficult words to help the A.I. choose the right one. Here’s a blog post explaining how to use this for custom vocabulary. I used an MLB draft video to illustrate how the glossary can help. And another blog post on WHY the A.I. needs help. More ‘word processor’ like text editor: A.I. isn’t perfect, even though it’s pretty close in many cases (usually 96-99% accurate with good audio). However, you can correct any mistake you find with the new text editor! It’s quick and easy to use because it works just like a word processor built into Premiere. Watch the tutorial on editing text in Transcriptive! Align English transcripts for free: If you already have a script, you can sync the text to your audio track at no cost. You’ll get all the benefits of the A.I. (per word timing, searchability, etc) without the cost. It’s a free way of making use of transcripts you already have. Watch the tutorial on syncing transcripts in Premiere! Adjust timing for words: If you’re editing text and correcting any errors the A.I. might have made it can result in the new words having timecode that doesn’t quite sync with the spoken dialog. This new feature lets you adjust the timecode for any word so it’s precisely aligned with the spoken word. Ability to save the transcript to any audio or video file: In TS 1.0 the transcript always got saved to the video file. Now you can save it to any file. This is very helpful if you’ve recorded the audio separately and want the transcript linked to that file. More options for exporting markers: You can set the duration of markers and control what text appears in them. Profanity filter: **** out words that might be a bit much for tender ears. More speaker management options: Getting speaker names correct can be critical. There are now more options to control how this feature works. Additional languages: Transcriptive now supports over 30 languages! Checks for duplicate transcripts: Reduces the likelihood a clip/sequence will get transcribed twice unnecessarily. Sometimes users will accidentally transcribe the same clip twice. This helps prevent that and save you money! Lock to prevent editing: This allows other people to view the transcript in Premiere or on Transcriptive.com and prevent them from accidentally making changes. Sync Transcript to Sequence: Often you’ll get the transcript before you make any edits. As you start cutting and moving things around, the transcript will no longer match the edit. This is a one-click way of regenerating the transcript to match the edit. Streamlined payment/account workflow: Access multiple
A.I. Speech-to-Text: How to make sure your data isn’t being used for training
We get a fair number of questions from Transcriptive users that are concerned the A.I. is going to use their data for training. First off, in the Transcriptive preferences, if you select ‘Delete transcription jobs from server’ your data is deleted immediately. This will delete everything from the A.I. service’s servers and from the Digital Anarchy servers. So that’s an easy way of making sure your data isn’t kept around and used for anything. However, generally speaking, the A.I. services don’t get more accurate with user submitted data. Partially because they aren’t getting the ‘positive’ or corrected transcript. When you edit your transcript we aren’t sending the corrections back to the A.I. (some services are doing this… e.g. if you correct YouTube’s captions, you’re training their A.I.) So the audio by itself isn’t that useful. What the A.I. needs to learn is the audio file, the original transcript AND the corrected transcript. So even if you don’t have the preference checked, it’s unlikely your audio file will be used for training. This is great if you’re concerned about security BUT it’s less great if you really WANT the A.I. to learn. For example, I don’t know how many videos I’ve submitted over the last 3 years saying ‘Digital Anarchy’. And still to this day I get: Dugal Accusatorial (seriously), Digital Ariki, and other weird stuff. A.I. is great when it works, but sometimes… it definitely does not work. And people want to put this into self-driving cars? Crazy talk right there. If you want to help the A.I. out, you can use the Speech-to-Text Glossary (click the link for a tutorial). This still won’t train the A.I., but if the A.I. is uncertain about a word, it’ll help it select the right one. How does the glossary work? The A.I. analyzes a word sound and then comes up with possible words for that sound. Each word gets a ‘confidence score’. The one with the highest score is the one you see in your transcript. In the case above, ‘Ariki’ might have had a confidence of .6 (out 0 to 1, so .6 is pretty low) and ‘Anarchy’ might have been .53. So my transcript showed Ariki. But if I’d put Anarchy into the Glossary, then the A.I. would have seen the low confidence score for Ariki and checked if the alternatives matched any glossary terms. So the Glossary can be very useful with proper names and the like. But, as mentioned, nothing you do in Transcriptive is training the A.I. The only thing we’re doing with your data is storing it and we’re not even doing that if you tell us not to. It’s possible that we will add the option in the future to submit training data to help train the A.I. But that’ll be a specific feature and you’ll need to intentionally upload that data.
