Skyhigh TV automates ‘logging’ of rushes using speech technology

Skyhigh TV is one of the Netherlands’ leading production companies and produces TV programs for both commercial and public broadcasters. The company mainly produces reality TV shows, docusoaps and documentaries.

What is logging? And why does it take so much time?

The key goal of any production is to find the best story possible. A crucial part of this discovery process is ‘logging’ of raw video footage (also called ‘rushes’). Loggers and transcribers usually have the task to go through the entire amount of raw video footage to transcribe conversations and to mark scenes for the final edit. By definition, there is always more video recorded than is used in the final edit.

It is the task of the Director or Edit Producer to finalize a final edit from hundreds of hours of video material. In order to accelerate this process, loggers and transcribers are used who type out all suitable fragments with their corresponding time codes.

“Normally we use people for this, who transcribe all conversations in the video material. For an interview of 2 hours, a logger usually takes up to 10 hours. That’s almost one and a half days of work’, says Sanne, Producer at Skyhigh TV.

In addition, it is an operational challenge to manage spotters, often students or trainees. If a whole batch of 10 hours of material has to be logged, 3-4 people are put to work who have to comply with deadlines and who have to spot according to certain standards. “This ‘people management’, of course, also takes a lot of time.”, says Sanne.

The scripts are then loaded into Avid or Premiere so that editors and directors can quickly access a selection of the right video material.

Skyhigh TV recently started working with Amberscript to automate the process as much as possible.

Amberscript automates almost the whole process

SkyHigh TV uses Amberscript to have videos transcribed using automatic speech recognition (ASR). The workflow changes. After production, video material is loaded into Amberscript und producers receive a complete transcript within only a few minutes. Amberscript automatically distinguishes between speakers and every word gets a timestamp.

In order to type out an interview of 2 hours, loggers now need around 20 minutes. Afterwards, some adjustments have to be made in some cases because speech recognition does not always work 100% accurate.

“For recordings with good sound quality like interviews it works incredibly well. I think many of our teams can’t think of working without Amberscript anymore – we literally save hundreds of hour of work. The trainees are of course very happy and our producers have much more time at our disposal. It’s a huge relief to spend less time distributing video files to different people and to monitor deadlines. In the end, this is time we can now use for creating high-quality TV programs.”

Apart from SkyHigh TV, Amberscript also provides services to filmmaker Rogier Kappers, read more about that here!