PxPixel
CEDAR DNS1000 Dynamic Noise Suppressor - ProSoundNetwork.com

CEDAR DNS1000 Dynamic Noise Suppressor

The diminutive CEDAR DNS1000 Dynamic Noise Suppressor may well create its own new product category. This impressive device allows the user to interactively define the shape, level and tonal characteristics of noise to be removed from final audio in real-time.
Author:
Publish date:

The diminutive CEDAR DNS1000 Dynamic Noise Suppressor may well create its own new product category. This impressive device allows the user to interactively define the shape, level and tonal characteristics of noise to be removed from final audio in real-time. It is fast, clean and puts an impressive amount of DSP at your fingertips.
Product PointsApplications: Post production, mastering, forensic audio

Key Features: Real-time dynamic noise suppression; AES/EBU and S/PDIF digital I/O; 40-bit internal resolution

Price: $5,895

Contact: CEDAR Audio USA, 207-773-2424 Web Site

Plus

+ Ease of use

+ Effective on a variety of noise types

+ Real-time operation

Minus

- Lacks attack and release control

The Score: Speed, functionality and ease-of-use combine to make the CEDAR DNS1000 a powerful tool.
Features

Sized between a Palm and a notebook PC, the DNS1000 ($5,895) is a dedicated digital signal processor offering 120 million floating point operations per second at a 40-bit internal resolution. I/O is digital only, with both AES and S/PDIF ports available on the back panel.

CEDAR has engineered a substantial amount of computing power into this physically unassuming product. An elegant user interface provides a broad range of precision control with only seven long-throw faders and seven illuminated buttons.

Starting at the top left are three pushbuttons for selecting which of the stereo digital input channels are to be processed. Selecting either Channel 1 or 2 applies processing to that channel only, allowing the remaining channel to pass through unaffected, while pressing stereo applies processing to both.

The next three buttons to the right are frequency range selectors, labeled "low," "mid" and "high." These switches can be used individually or in combination with each other to define a wide variety of frequency ranges for the control faders below.

These are seven long-throw control faders. The far leftmost, labeled "level," is adjusted to set the threshold point below which the signal is deemed to be unwanted. The level fader is calibrated from -80 dB to 0 dB.

The six faders to the right, labeled "gain," are used to determine the amount of gain change for signals falling below the threshold. These are calibrated as -24 dB to +6 dB, which suggests that any frequency band may be amplified as well as reduced - and this is indeed the case.

Although there are only six faders, the audio is divided into many more bands than this (CEDAR does not specify the exact number) and processing is applied to each individually. This makes the DNS1000 more accurate in its identification and removal of noise than a simple, six-band processor.

A red and a green LED are positioned above each of the gain control faders. These provide a visual indication of the average activity taking place within each of the six visible "bands" (as defined by the faders), and prove to be extremely useful in tuning the unit for the best results.

An ingenious feature of the DNS1000 is the way in which the range selector buttons define the frequencies to be affected by the six control faders. These buttons are pressed individually or in combination with each other to produce six possible processing ranges. The frequency bands controlled by the faders are evenly distributed across the selected range, effectively zooming the control range in or out.

Pressing "low" defines the processing range as between 20 Hz and 400 Hz. "Mid" defines the range as 200 Hz to 6 kHz, and "high" as 4 kHz to 18 kHz. The combinations are "low+mid" for 20 Hz to 6 kHz, "mid+high" for 200 Hz to 18 kHz, and "low+high" for a full range of 20 Hz to 18 kHz. This clever arrangement allows the operator to work in broad strokes or with finer precision. Finally, there is the all-important bypass button for before and after comparisons or the removal of the device from the signal path.

The manual is concise and to the point, and includes case studies of suppressing traffic noise, ambient sound, tape hiss and excessive reverberation. The focus of the unit is squarely on post production dialogue processing, but the flexibility of the control interface permits live application as well.

The DNS1000 was used to enhance the intelligibility of Tom Hanks' dialogue over the noise of the surf in Castaway and the unit is also finding application in forensics.

In Use

My personal experience with CEDAR products has been in the restoration of music recordings, where the company's offerings have traditionally set the gold standard. In contrast to a declicker or decrackler, a dynamic filter such as this is most effective at reducing continuous noise. Hums, buzzes, rumble and hiss are more conducive to removal than more complex disturbances, such as automobile engine noise or natural reverberation. The more textured the noise, the harder it is to distinguish from the program to be preserved.

Still, with good quality dynamic filtering as provided by the DNS1000, improvements can be achieved even with the more difficult noise sources. Efficacy of the device depends largely on the skill of the user in identifying the character of the noise and in knowing where to draw the line of diminishing returns. This is the line beyond which the desired signal suffers from too much truncation or phase shift.

CEDAR's well-conceived provision for up to 6 dB of gain in each band for signals above the threshold gives further flexibility in separating the wheat from the chaff, and lends itself to creative dynamic equalization as well.

To prepare for my DNS1000 trials, I made some voice recordings in the proximity of electric motors, in a reverberant stairwell and on a busy New York City street. Although it was not necessarily designed with this in mind, I also threw some old cassette dubs at the device to test its mettle with a musical signal. An improvement was possible in every case, and I was especially surprised at the results of the stairwell test. The voice did appear to come closer after treatment.

Street noise proved to be the most challenging because of the unpredictable and complex nature of the ambience, but enough selectivity was possible to make an appreciable difference. Cassette hiss was effectively reduced by selecting the mid+high range and by dialing back only the leftmost faders. This was suggested in the manual as a means of retaining more of a sense of air in the resulting audio.

With as much control as the device offers, I wish CEDAR included provisions to adjust overall processing attack and release time.

Summary

The DNS1000 offers a unique blend of speed, functionality and ease of use, packaged in an impressively small footprint. The audio quality is excellent. While there are software programs and customized PC hybrids offering spectral-based noise reduction, these can cost well into five figures and tie up an entire desktop computer. Like CEDAR products in general, the DNS1000 is not inexpensive, but considering what it does and how quickly it does it, the device could soon pay for itself in a post production environment where the kinds of problems it addresses are regular occurrences.