My latest project, the CORDA SOUL, has raised quite a lot of interest. Not only because of its superior sound quality but also because of the many options to optimize sound in the digital domain. The built-in DSP (digital signal prozessor) allowed to implement tone controls, notchfilter and crossfeed with a precision that easily outclasses any implementation in the analog domain.
The SOUL as a whole is a very sophisticated design and thereby unfortunately also is quite expensive. And many people already do have a DAC/headamp combination at home. Therefore there were many requests to built/design a stand-alone DSP that offers all the digital options of the SOUL.
I have been working on this processor for quite a while now and as it happens, various new ideas popped up. The current status of the design is presented here. As there is a lot of computing going on inside this project is named the COUNT(RY).
June 10. 2019
The front of the COUNTRY has a large number of control elements. Of course a more slick design could have been made with a LCD-screen, programming buttons, and a menu-structure to set all parameters, but the advantage of using separate control elements is, that one can check all settings at one glance.
The COUNTRY also has some functionality not found on the SOUL.
The left lower dial allows to activate a psychoacoustic bass-enhancer. Quite some years ago I developed a device that makes signals below 60 Hz audible in loudspeakers and headphones that normally, by their mere physical construction, are not able to reproduce these frequencies. This design was originally published at Headiwze, the site of the late Chu Moy, and now can be found at:
If interested please read carefully. It's not an easy read but an interesting one. Basically the bass-enhancer monitors all the signals below 60 Hz, calculates their harmonics, and adds these harmonics to the original signal. If properly done, these harmonics fool our brains in that they suggest the presence of the lower fundamental signal.
The original design was done using all-analog circuitry. However, with the COUNTRY it was very easy to implement the algorithm in the digital domain. To let you experience the effect of the bass enhancement I've prepared an audio-file.
The file presents 10 six-seconds sections separated by a 1 second pause. Take your headphone and give the file a listen by clicking the button below. You will hear:
1. A pure 100 Hz sinus signal
2. A 100 Hz sinus signal with bass-enhancement.
3. A pure 70 Hz sinus signal
4. A 70 Hz sinus signal with bass-enhancement.
5. A pure 50 Hz sinus signal
6. A 50 Hz sinus signal with bass-enhancement.
7. A pure 40 Hz sinus signal
8. A 40 Hz sinus signal with bass-enhancement.
9. A pure 34 Hz sinus signal
10. A 34 Hz sinus signal with bass-enhancement.
With a decent headphone it will not be difficult to hear the high frequency pure sinus signals. However, at 40 Hz and at 34 Hz the level of sensation will be very low. However, not so with the bass-enhancement engaged. Very suddenly these signal components become very audible.
Granted, the bass-enhancement does "colour" the sound. It is not the real thing and there is no visceral sensation. Yet it can be very helpful to improve the level of realism when we hear a recording that has deep frequency components.
Please compare the following two tracks:
I will not tell you which is the original recording and which uses the bass-enhancement. The effect of the bass-enhancement is rather subtle but you should be able to detect its effect. Please let me know your impressions and observations. Do you feel the bass-enhancement to be an interesting/valuable addition to the COUNTRY design?
The 8 slide-potentiometers each have a LED at the heart of the control-knob. These LED's are used to indicate sampling frequency (32, 44.1, 48, x2, x4), a straigh frequency response (with all controls of the equalizer set to the same position), -6/-12dB settings of the volume control ,as well as the ON-state of the device.
Especially with computer audio the indication of the sampling frequency is a great help to check the software settings.
Next to the crossfeed dial a slide-potentiometer allows to set the volume in the digital domain. Thus your DAC can be used as a preamplifier with the volume set by the COUNTRY. The output accuracy of the COUNTRY is 24 bit, so signal losses are minimal.
At the middle of the front a seven band equalizer is found. Each slide potentiometer allows to change the amplitude of its corresponding frequency band by +/- 6 dB (1 dB steps).
Central frequencies of the bands: 64, 160, 400, 1000, 2500, 6200, and 16000 Hz.
With the rightmost potentiometer set to its lowest position, the COUNTRY applies a de-emphasis filter to the digital signal. It can be used when a recording is known to have emphasis.
In its current design the COUNTRY has three digital-inputs. A coaxial RCA-input, an USB-input, and an optical TOSLINK-input.
The USB-input allows sampling frequencies of 32, 44.1, 48, 88.2, 96, and 192 kHz. Resolutions of 16 or 24 bit.
The coaxial and optical inputs also allow sampling frequency of 64, 128, and 176.4 kHz. Resolutions 16, 20, or 24 bit.
With the SOUL all digital processing is done at the native sampling frequency of the recording. However, with the COUNTRY Internally all input signals are resampled before processing. This makes the programming more easy and lowers the computing power required.
Originally my goal was to use an internal sampling frequency of 192 kHz. However, some of the algorithms (especially reverberation) are rather complex and probably do not "fit" within the limitations of the computing power. If so, it may be necessary to use a sampling frequency of 96 kHz instead or, as a more luxurious solution, use two DSP-chips instead of one.
There are two digital outputs: One coaxial RCA-output and one optical TOSLINK-output.
The input signals are resampled by the ADAU1463 DSP-chip by using an extremely high-quality TCXO as a reference timer.
Next the digital audio signal is internally processed with 64-bit accuracy.
The USB-input uses the CM6631 from C-Media to receive data in asynchronous mode from your PC. The use of extremely high-quality TCXOs guarantee jitter-free operation.
DSD-signals and 384 kHz signals can not be processed by the DSP-chip and therefore are not accepted. I'm sorry for that!
Functional prototypes have been finished and currently are used for fine-tuning!
If interested in this very special project, please do not hesitate and drop me a mail.
The lower right dial controls crossfeed.
In the upper (neutral) position the output is plain stereo.
Turning the dial clockwise activates headphone crossfeed (7 levels)
Turning the dial anti-clockwise activates loudspeaker crossfeed (7 levels). This widens the soundstage of loudspeakers.
The upper right dial has two different functions:
- In stereo or headphone crossfeed mode the dial controls the central frequency of a 6dB, Q=2 notchfilter (see picture below). Many headphones do have a distinct resonance frequency due to the entrapped air-volume between driver and ear. Proper application of the notch filter will lower the energy at the resonance frequency and this can result in a remarkable improvement of quality of sound.
- With loudspeaker crossfeed activated this dial allows to optimize the delay time between the direct signal and the crossfeed signal. The optimal delay time strongly depends on the positions (angles) of the loudspeakers.
Of course if you have some additional ideas of algorithms that should/could be implemented into the COUNTRY, please let me know.
For myself I have been working on an algorithm to improve the timing of a loudspeaker. Due to the crossover network different frequency bands do have different polarity and are (partly) delayed in time. This has major consequences for the impulse and step response of the loudspeaker.
The upper figure to the left gives you the step response of my own loudspeakers measured at close range. It's a 3-way system with bass reflex port and this clearly shows in the response. Believe it or not, the frequency response is pretty flat, but you would never guess from this measurement.
When digitally compensated for the time delays and polarity changes, the response clears up and now easily can be recognized as the result of a signal step (lower figure). Sure, after the step the signal decays, but that's merely a result of the low-end limitations of the speaker. And believe it or not, the frequency response is totally unchanged !
Sonically the time compensation translates in a more precise sounstage and a tighter control in the bass.
I do not plan to implement this algorithm into the final version of the COUNTRY. Every loudspeaker model needs different compensation and the algorithm occupies most of the computing power of the DSP-chip.
The left upper dial allows to add reverberation. Especially with recordings made in relatively small venues, often very little natural reverberation is added by the sound engineer because our own living room already adds some of its own when listening to loudspeakers. However, when using headphones this extra reverberation is missing and sound can be a little bit dry. Adding some (high quality) artificial reverberation by the DSP then definitely can have a positive effect. Right now this feature only needs some fine tuning.
As an example listen to the three tracks to the right. It's a real pop-classic. The first file is the original track. The second and third file have different levels of reverb. Again, your impressions are much welcome.