Issue link: http://digital.copcomm.com/i/830490
C A S Q U A R T E R L Y S P R I N G 2 0 1 7 43 launching the DNS 1000 back in September 2001, and we soon found that location sound recordists and outside broad- cast teams (what in the US you'd call remote teams) were using the DNS 1000 in the way in which we now envisage the DNS 2000 is most often used. What were the obstacles associated with making it work in the field? The huge difference between the manually controlled DNS products and the most recent ones is the development of the Learn Algorithm, which enables the new models to determine the noise content of the signal automatically, even while the speech is going on. And it wasn't until that mathematics was developed that, eventually, a version showed very promising results. Once that (Learn Algorithm) had been developed, it was then a question of making it as computationally efficient as possible and making it possible to control with the mini- mum number of controls so that it was very simple and intuitive and quick to use in the field. Although the algo- rithm is a DNS process, the way the audio is handled (is) completely different. I would imagine we started developing the Learn Algorithm sometime in the mid-2000s, because we launched the first product with it in 2012. Understanding that some mixers are reluctant to do a lot of processing in the field, do you find instances where people use it more to audition the possibility of noise reduction for post, or recording processed and unprocessed tracks simul- taneously? All of the above. The DNS 2 has the ability to take a single channel of audio and output both the raw audio and the pro- cessed audio simultaneously, so that deals with those people who are, understandably, cautious that what they're hearing in the field may not be the best possible and that they want the raw audio for post. It's also used in very much the same way as a confidence test, and one will often find that the lower the budget, the more frequently the audio processed in the field is used (in the final product). And if you can cut your post time by a significant percentage, and your post costs by a significant percentage, and you don't expect the audio to improve noticeably, then the audio that's processed in the field may well be as good as it gets. How is this changing production across the board? From people who are now live broadcasting in very difficult environments that previously might not have been possible. Particularly, sports broadcasters and news broadcasters are now able to go live to air intelligibly and with the background noise suppressed to an adequate degree that you still have the atmos of what's going on but without it drowning out what you want to hear which is, of course, the reporter or anchorman. For film production, particularly in the low-cost arena, it's made a huge difference because some low-cost film and TV productions that possibly didn't even have the budget for any kind of serious cleanup in post, now the audio that's being provided to the mixers may already be suitable for the final production. Has noise reduction reached its peak? A most resounding no! It's come on a huge long way in 30 years, but it's still got a very long way to go, and there are many recordings and many live situations where current noise reduction remains inadequate. I think there's a lot of research to be done, a lot of new products, and a lot of exciting new algorithms still to be developed. Any closing words? We are always hugely grateful for recognition from the Cinema Audio Society and we value the recognition of the CAS very highly. Congratulations to CEDAR for its groundbreaking field noise reduction system!