Following a request for a recommendation on Twitter for speech to text transcription options I waded in perhaps a little too eagerly before reading the context of the post request for recommendation for responding.
Transcription from audio for a text record and extracting text for uploading as caption are similar by not the same.
I immediately recommend Headliner. I love services that evolve and extend their usefulness. I first used Headliner to generate overlay audio waveforms to present podcasts for viewing when uploaded to video services.
Last year I worked on a spoken word project requiring captions to accentuate particular words and phrases. It was also audio and needed audio visualisation. I went to Headliner to generate a selection of waveforms, Headliner calls these audiograms, to demonstrate the options and discovered the transcription for captions feature. For transparency I haven't as yet fully pushed Headliners captioning feature to the max but I wanted to tell this story.
NOTE: the waveforms are generated using Headliner the words were created using titles in iMovie as they needed to be closely match to the intonation of the narrators words and inflections.
After an exchange on Twitter with Headliner we scheduled a video call for me to share my Headliner experience.
Using the scheduling power of Calendly I had a great chat with Mike Sands from development at Headliner.
The generation of transcripts I’ll deal with tomorrow but for closed captions which is where headliner can facilitate and excel as a player in the field for affordable rich featured captioning I wanted to share it with you. The purpose of this blog is really to highlight the issue below that having a conversation directly with a platform's development team can facilitate.
Headliner also takes SRT file and can export SRT files. SRT are text format files that time code each line for a duration to appear on screen.
Here is an example of a few lines for an SRT file
00:00:47.769 --> 00:00:51.002
00:00:52.584 --> 00:00:54.030
00:00:54.409 --> 00:00:56.365
00:00:58.498 --> 00:00:59.883
00:01:02.233 --> 00:01:04.216
You haven't seen a potato anywhere, have you?
00:01:06.389 --> 00:01:07.611
00:01:08.381 --> 00:01:09.528
He ran off.
SRT buffering was my 'you could be the golden goose' opportunity. Completed videos are provided from artists for live screening or on demand viewing. But… the festival may wish to insert a pre show bumper to brand the experience. That add time to the start of the video file and thereby changes the time code for the start of the captions. You can see this from the time stamps in the example from Shona Reppe’s Potato Needs a Bath digital experience - https://www.shonareppe.co.uk/potato-digital.html
For the 2021 SparkArts Festival a bumper lasting just under 47 seconds was inserted at the start. To off set the start of the captions was a trial by fire. Every time code through the entire SRT file needs adjusting. The example shows less than a minute of the show in lasted 30 mins.
This conversation revealing this step in a process is an opportunity. I had to download and install sub shifter could Headliner keep me in the platform.
Lets hope this small hurdle but useful addition gets actioned.
If you want to try out headliner here is a link for a free trial. After you can use the free features or pay for a month or for a year.