Don't see the answer you need? Contact us with this Request Form.
PLEASE report any issues you run into to firstname.lastname@example.org.
Make sure you’re using Speechmatics. It has much higher accuracy than Watson. Also, background noise and unclear speaking can make it harder for the speech services to figure out what’s being said. The less noise you have and the better
your talent enunciates words, the higher the accuracy will be. Make sure your talent is mic’d, speaks clearly and any background noise is minimized. Talent that has a heavy accent or is off mic will get transcribed with less accuracy (sometimes significantly less) than talent that is mic’d and well spoken. Regardless, Speechmatics is the better Speech Service, especially with recordings that have a lot of background noise. We highly recommend you use Speechmatics if you’re doing captions or subtitles.
If you’ve already gotten a transcript for that sequence, you need to select Load Transcripts from the Main Menu. This will load the transcript for the sequence you have selected. Premiere is not that great about telling Transcriptive what’s going
on, it’s a bit like pulling Adobe’s teeth. Not fun. So if you switch to a different sequence you need to tell Transcriptive that you’ve done so by selecting Load Transcripts. Of course, if you’ve never generated a transcript for that sequence, you’ll need to click the Transcribe button and have it transcribed.
It does work! Use Control+Z for Undo and Control+Shift+Z for Redo. A Premiere panel can’t use the regular Command+Z as a normal application would, unfortunately.
In order for Transcriptive to work, you have to click on the Transcriptive panel. If you just select your sequence and hit play, Transcriptive won’t do anything.
This is a limitation of Premiere. You have to select the Transcriptive panel and hit play from within Transcriptive otherwise Transcriptive won’t have any idea what Premiere is doing. Transcriptive is technically a separate app running within Premiere. It will communicate with Premiere but not vice versa. So control Premiere from within Transcriptive, it won’t work otherwise.
You enter Edit Mode by clicking "Enter", which allows you to edit text as you would in a word processor. You exit Edit Mode by clicking "Esc".
It's a limitation of Watson. They only transcribe 100mb files, so it's limited to however long of audio we can squeeze into 100mb. That translates into about 70 minutes of audio. Watson recently started allowing us to upload mp3 files.
Transcriptive 1.0.1 will include this and allow you to send up to 3hours and 30 minutes to Watson.
Speechmatics has no limit and will take as long of an audio file as you can throw at it.
To create closed captions in Premiere, you need to export the transcript as SMPTE-TT and import that into Premiere. Premiere’s Caption panel will see that as 608 or 708 captions.
The caption file starts when the first word is spoken. So if your talent doesn’t start speaking immediately, you don’t want the start of the caption file to be at 00:00. Move the start of the caption file to line up with where the talent starts speaking and everything will be in sync.
Although the batch transcribing is successful, Premiere does not refresh instantly for every clip’s speech analysis to show within the metadata panel. It can take some time before it shows. You can also reimport or replace the file in your project and the speech analysis will show instantly.