Saturday, February 28, 2015

A Better Arrangement of the Sound Transformer Horns

This is a continuation of The Sound Transformer - An Early Prototype.

With all three horn drivers in the same plane, a listener to this system is about seven feet further away from the midrange and tweeter horns than from the bass horn.  This location of the horns does achieve reasonably accurate "time coherence", i.e., the sound from each of the three speakers arrives at about the same time.  However, as a result the system "sounds" bass heavy unless you move away from the system.  That is not necessarily a problem if you enjoy strong bass.

Another consideration is the amount of space occupied by an eight-foot long straight bass horn even in a relatively large room in a house.  As it turns out, since 1934 when all of the early research on this was done at Bell Laboratories, it is now possible, with a modest investment in DSP hardware, to achieve "time coherence" between the woofer and the mid and treble speakers by introducing a time delay in the signal path.  Digital Signal Processing (DSP) makes it possible to do this in "real time" during playback.  With DSP time-alignment correction, it would be possible to place the mouth or delivery end of the midrange/treble and bass horns in the same plane and apply a delay to the midrange/treble horn signal.

To check this out, I mixed a recording in Audacity, the excellent open-source audio processing software, in the following way.  I only have one of the speakers so I have been working with monaural tracks.  In Audacity, it is possible to load a stereo track and convert it to a single monaural track.  After that, you can run a low-pass filter over the track and save that filtered track to a monaural track. I named this low pass track "track-lp-400Hz".  Then you can "undo" the filter and run a high pass filter on the same original monaural track and save that as "track-hp-400Hz".

Finally, you can load both the lp and hp tracks into a new project and assign one to the left and one to the right channel.  Now you have essentially built an "active crossover" into the program source file.  The last step is to use the "time shift" function in Audacity to slide the midrange/treble signal about 6 milliseconds to the right on the time scale relative to the bass signal.  This step builds "time-alignment" into the system.  The speaker horn arrangement looks like this:

Sound Transformer With DSP Time-Alignment
Sound Transformer With DSP Time-Alignment


Sound Transformer With DSP Time-Alignment
Sound Transformer With DSP Time-Alignment


Sound Transformer With DSP Time-Alignment
Sound Transformer With DSP Time-Alignment
Now, the balance among the three horns is really quite good no matter where you are in the room.  Also, with this arrangement, it would be possible to build the whole assembly behind a wall so that it would not obstruct the listening room.  By the way, no passive crossover is necessary in this hook-up.  The midrange and bass horn drivers are connected directly to the right and left outputs of the Lepai 2020A amplifier.  Only the tweeter horn is protected by a series capacitor.  In addition, the tweeter horn attenuator network is still in place.  As soon as I have more on DSP time-alignment, as applied to this system, I will post here.

No comments:

Post a Comment