Join Maxim Jago for an in-depth discussion in this video Improving speech-to-text analysis, part of Video Production with Creative Suite 6.
You might think of Adobe Story, which is what I have on screen here, as being exclusively for script writing, scheduling and reporting. And various other pre-production tasks. But it actually has a very valuable service it can provide for Premiere Pro CS6. Premiere Pro can analyze the audio in your clips, and convert it into text, and then you can edit based on that text. And Story can provide that text. Now, you can just give Premiere Pro a text document.
But if you give Premiere Pro an ASTX file, this is the file exchange format you can get from Adobe Story. Then you can get use for things like character names as well. The process is very very simple. I've got a script here which is actually a transcription of a little bit of dialog by me as I'm describing some shots from some virals that I produced in advance of my feature film or fierce writing. So, here is the transcription. And just note that this is a transcription, this is not an original fiction script. And of course, if you're working with fiction, it's very, very common for the dialog to change, not just during rehearsal, but even on set. And it's not always possible for your script supervisor to catch those little changes, some of which might well end up in the edit anyway. So, here I am with my transcription.
and I'm going to go to the File menu and I'm going to choose not save to disk which will create an ST DOC file. I've put that in with the other assets for you anyway, so you can open this up and try this for yourself. But also export as, and I'm going to choose Adobe Story Interchange format ASTX. I'm going to save this, so here you can see we've got a folder, we got one. There's our Script folder, and inside there I'm going to save this as Directors Commentary.ASTX. Notice the original is ST DOC. And I'll save that, nothing much is going to happen. I'm going to toggle over to Premiere Pro.
And here, I've got this project, improving speech to text. And this just has the screen capture images on it. This is an MP4 file I've converted, and you can hear me speaking. Some of the ways we worked on the viral media that we shot in prepara, and there it is. So, what we want is to go to our Metadata panel. And if you don't have this on screen, you can just go to the Window menu and chose Metadata. And in this panel, you can see if I select the clip in the bin, you can see right at the bottom under Analysis text. You see there's nothing there at all.
Now, if I want to I can click Analyze at the bottom of the Metadata panel, and I can just set this off conducting its analysis, and it will do okay. But I want to show you another way. Now, you can do this in the bin by right-clicking on the clip and choosing Analyze Content, or you can do it here in the Metadata panel. If I click on Reference Script and I choose Add, and then browse to that, ASTX file, here we go. Notice, that there is an ST DOC file in this folder, but it's not being displayed.
Because the ST DOC files have additional formatting information that we just don't want or need for this process. The ASTX is the one. But remember, the ASTX files don't have that formatting information, so be sure to store both if you're looking to make a local backup of your script. I'm going to select this and click open and here you can see the dialog taken from the script. You see at the top I can choose that this is English. Perhaps, a little bit optimistic saying it'll be the same accent from so many part of the UK, but it seems to work out most of the time. And up and the top here, this is the tick box I want to draw your attention to. Script text matches recorded dialog.
What this means is I'm going to take this box and the analysis engine, which is actually not Premiere Pro. It's actually the meter encoder that does the analysis. It's going to presume that exactly these words and no other words are spoken. And this is important because it's going to give us an absolutely perfect speech to text analysis, provided that this is the actual text that's spoken. So, you can see why this is particularly good for transcribed audio. Although there's no automatic way of generating subtitles from this content, you can use the cue points that will be generated automatically if you send this media over to after effects.
And those cue points can be turned into subtitles effectively using expressions. So, just click OK here. I'm going to have this on high quality. I'm going to make sure Identify Speakers is ticked so I get the name of the speaker, and I'm going to click OK. Up comes the Meter Encoder. The Meter Encoder begins its analysis. Now, if I re-size this a little so you can see what's going on, we get the little progress bar at the bottom. It doesn't take too long. Of course, it's not outputting any video, so the Preview panel here is blank. That's now completed.
And if I toggle back to Premiere Pro, I'm just pressing Alt+Tab here, that will be Cmd+Tab on Mac OS. And here it is, there is the text. Exactly what I said when recording this video. And you can now click on the word, and notice the playhead in Premiere Pro is jumping to the word. So now I can say lets go from timeline conversion of this shot on the timeline. You can see exactly, and I can actually cut the pictures based on the words that are being spoken. This is the speech to text analysis feature in Premiere Pro.
But if you like perfected by using the data provided by Adobe Story. A massive time saver for the editor if you can go through that preparation time before you begin.
- Improving speech-to-text analysis with Story
- Organizing projects in Prelude
- Batch renaming with Bridge
- Preparing images for video in Illustrator
- Working with Photoshop files in Premiere Pro
- Round-tripping a soundtrack from Premiere Pro to Audition and back again
- Preparing content for After Effects in Premiere Pro
- Sending work from Premiere Pro to SpeedGrade
- Using Dynamic Link to share sequences between Premiere Pro and Encore
- Using the Media Encoder to output from After Effects