|PRODUCT: Cedar DNS3000
- Hardware controller with automated faders
- Near zero latency
- Mac or PC
For many audio professionals, Cedar tools have long been the top choice for audio restoration. The DNS3000 is the latest generation of Cedar's Dialogue Noise Suppression technology, combining the control surface of the original DNS1000 with the Digidesign Pro Tools integration of the DNS2000.
Additional features include on-board scenes with a recall system, automation to timecode, moving faders and sample rates up to 96kHz. It is not a plug-in, but a hardware device that provides the processing power and audio I/O. The result is near zero latency to ensure that there is no loss of lipsync.
Using the DNS3000 made me rethink how I do dialogue restoration. Once a frequency range is selected, it starts with the maximum amount of noise suppression with all the faders down. Then a combination of faders are selectively raised, or not, to restore the clean audio minus the noise.
For example, I needed to remove background noise and wind gusts from production dialogue on a commercial for PreEmptive Meds, Inc. The actors were shot outside, on location in Florida as a hurricane was brewing, so the amount of wind to remove was quite heavy. First, I selected the frequency to be removed. The choices are similar to an EQ: high, mid, low, high and mid, mid and low, and full.
Once selected, I pulled all the faders down for the maximum amount of noise removal and then I selectively raised the faders to add back the clean sound. The workflow was a little backwards to my mindset, but it worked very well, even though there was a high amount of guess work involved. I was not really sure what frequency range the wind was at, so I needed to select each range separately and repeat the steps to see which one yielded the best results. Since the commercial used multiple actors at multiple locations and cut between multiple takes, I needed to do this for every single line spoken for two commercials and two eight-minute Webisodes. Needless to say, it was labor intensive.
The DNS3000 has a snapshot/recall function to remember your restoration settings (like actor number one needs setting X, while actor number two needs setting Y). I found the snapshot feature to be much too cumbersome and time consuming to use. Therefore, I took my own approach of saving my work by simply re-recording the clean dialogue back into Pro Tools onto new tracks. Cedar does not specify to work this way, but I found it to be a more realistic approach for quickly and permanently saving my numerous restoration settings and the hours worth of work that went along with it. Once completed, my approach gave me the ability to mix in a different studio that did not have the Cedar box, since we are a multi-room facility.
Even though it has a software interface that looks and works like a plug-in, it is not a plug-in. The software does not process audio in anyway. It only tells the hardware how to process the audio, which is routed in and out of Pro Tools. This is similar to an insert on a console when patching outboard gear. Therefore, anything done on the software is not saved in anyway until you tell the hardware to take a snapshot/recall it, or to re-record it back into your DAW, like I did.
Having the processing power outside the host computer has two main audible benefits: First, is the near zero latency, which is less than 10 samples according to their documentation. The result keeps the proper lipsync to any processed audio. Second, and more importantly, it allows more processing to be done with less audible by-products. It was so nice not to hear that strange digital reverb-y sound leftover when using the competition's restoration plug-ins. Cedar focuses on one objective — cleaner audio — and it shows.
The biggest drawback I found with the DNS3000 was that it forced me to guess where the unwanted audio frequencies were. Anyone familiar with unwanted background noise knows that often the sound to be removed occupies most of the sound spectrum. Rarely is it a simple matter of removing one singular frequency that is not found anywhere else in the dialogue.
Since I had a hard time guessing, I really wished for some kind of noise sampling, or learning function, that would pinpoint the range of noise for me. Sure, as the manual said, I got better at identifying the noise the more I did it, but I missed the instantaneous precision that some of the competitors offer with their plug-ins.
Another drawback was that it was not easy to use or intuitive with its functions. I found it to be a slower and more difficult task to clean dialogue using this device, even though it gave me better results. Even with extra time spent, I found that certain things, like removing the director's voice mistakenly recorded during a take, were impossible to do. Bus brakes screeching, background music, and water fountains also proved to be impossible to remove completely.
Cedar is as good as it gets when it comes to dialogue restoration. The DNS3000 is a top-notch product, but is not 100 percent effective. For those studios that do dialogue restoration full time, I can see them getting their money's worth. But, for those studios that do occasional clean-up, there are many cost effective restoration plug-ins to choose from.