Design Article
Pro audio techniques applied in mobile devices to improve sound quality
Earl Schreyer, Fairchild Semiconductor
7/26/2011 10:11 PM EDT
Figure 2 below shows an example of compression response curves for 2:1, 4:1, and infinity:1 (limiting). This demonstrates that, below the DRC threshold, the output vs. input has a fixed gain response. After the DRC threshold is crossed, the gain changes proportional to how far above the DRC threshold the original signal would have gone. Following that, it becomes clear what amount of compression will be applied for the given compression ratios as described above.

The engineer can then increase the system gain to bring out the lower levels of the audio, while compressing the potentially distorting or over loading signals. You may hear the term "normalizing" the gain, where the engineer re-adjusts the system gain so that the highest levels (peaks) are synchronized to the same level as they were without the effect enabled, but now the lower amplitude signals below the DRC threshold will be boosted.
As compression ratios are further increased (above more than 10:1 for example), the effect starts to become what is called a 'limiter'. In this case, the automatic gain control will continue to decrease the gain such that the output amplitude will no longer be allowed to increase any further after the DRC threshold is crossed. In mobile devices it can be useful to implement both techniques together.
You could think of the compression region as a gradual approach to limiting, and if a higher final limiting level is reached, the limiter will kick in and prevent the signal from increasing any further at the output. Limiting can be useful in mobile devices to track the battery voltage. For example, if the limiting threshold tracks (is proportional to) the battery voltage, the limiter level will decrease along with the battery voltage to prevent distortion as the battery voltage naturally decays.
With Figure 3 you can see an example of the two effects working together for a combined response. In addition, with this setting, the initial gain is 24dB, and when the DRC threshold is crossed the circuit response moves to a 2:1 compression region. Finally, if the output signal would have peaks over 90% of the battery voltage, the output gain will be reduced further so that the output signal level cannot go above the limiter level. Then this limiter level would continue to move down as the battery voltage decreases since its value is always 0.9 * Vbatt.

Compressors are also often supplied with attack and release times to control the speed at which the circuit operates on the signal. Usually it's desirable to attack down on the peaks rather quickly as not to overload and distort the speakers, and then allow the effect to release slowly as the envelope of the audio signal stays below the threshold level.
Figure 4 shows an example of attack and release times. The signal is shown as a pulse to clearly illustrate the attack and release times. Again for illustration purposes, the attack time is shown just slightly faster than the release time.

These controls can be used to tailor the effect for the given application to optimize its usefulness. For example, a phone conversation may have different settings than playing music, or different instruments in a studio recording environment may have different settings to tailor the effect to that given instrument.
Since Fairchild audio subsystems have high-resolution digital volume controls, the gain can dynamically be controlled within the device to approximate the ideal compression ratio curves and limiting response when threshold levels are sensed internally to the device. These devices have on-board trimmed oscillators, so a very accurate time base can be applied to create programmable attack and release times for the DRC and limiting features.
New Fairchild mobile audio subsystem ICs on the market and soon-to-be-released products incorporate both fixed THD limiting as well as battery voltage scaled limiting, while others include fixed THD limiting, battery scaled limiting as well as dynamic range compression. Furthermore, boosted stand-alone class-D speaker amps from Fairchild incorporate limiting as well.
With the small low-profile speakers in today's mobile devices, these traditional Pro Audio techniques can be used to raise the lower level audio signals in a conversation to aid in communication (so you don't miss what the person is saying). Taking advantage of the multimedia entertainment features, compression and limiting can also be applied to maximize the audio level for listening pleasure (for a movie or video for example) while not allowing the louder dynamics to reach an unpleasant and potentially speaker damaging level.
Author bio:
Earl Schreyer has 20 years experience in designing integrated circuits. He began his career with Fairchild Semiconductor in 1999 joining the analog & mixed signal products group, building on analog and high speed ADC work in the application of video/graphics digitizers and continues time video filter drivers. He has a passion for audio design and has recently become a key member in the new mobile audio group.
Earl holds both a Bachelors and Masters Degree in Electronics Engineering from Arizona State University. In his spare time, he likes to design and build hand-wired, vacuum tube guitar amplifiers.
For more articles like this and others related to audio design, visit Audio Designline and/or subscribe to the monthly Audio newsletter (free registration).


kdboyce
7/26/2011 10:52 PM EDT
The audio quality problem with small speakers in portable devices is being attacked on several fronts, with DRC, AGC, and limiting being among the 'classical' ways to help the situation. Active multi-band EQ is showing up more often, and OEM's are tending to place more emphasis on acoustics design to help counter some of the speaker deficiencies. Even psycho-acoustics methods (tricking your ears and brain) are being tried with some measure of success.
Sign in to Reply
gdavis9999
7/27/2011 1:31 PM EDT
Boyce, good comment. Here at Fairchild, we believe that one of the key attributes of DRC (dynamic range compression) is to achieve higher SPL (sound pressure level) for louder sound. The advantage is you do not have to use the “brute force” method of driving more current into the speaker element to gain some added perceived volume. Although more power is ultimately needed to make a large change to achieve higher output volumes, we see DRC as a supplement to that, without the downsides of added current such as higher power dissipation and possible speaker damage.
We will issue a major audio announcement soon that includes our first audio ICs optimized for mobile and tablet applications. We also have multiple devices currently in design that will continue to provide value in “making small speakers sound louder and better™”. This includes psycho-acoustic approaches. Thanks again for your comments, they are always welcome.
Greg Davis
Senior Marketing Manager
Fairchild Mobile Audio Products
Sign in to Reply
kdboyce
7/28/2011 4:04 PM EDT
Greg, your point is correct. Thank you for further information on Fairchild's activities. I am sure the market will respond well to solutions that work, as they are sorely needed.
Sign in to Reply
Dr. HD
7/28/2011 10:23 AM EDT
I think it's important to be clear that "louder" absolutely does *not* always equal "better", but this mistaken belief in some quarters has lead to the "loudness war" - sound recordings have copious amounts of DRC & limiting applied to them, robbing them of all dynamics and making them sound worse (see this excellent youtube video detailing the problem: http://www.youtube.com/watch?v=3Gmex_4hreQ, further info can be found here: http://en.wikipedia.org/wiki/Loudness_war and here: http://turnmeup.org/). The problem of applying DRC & limiting during mastering is that there's no lossless way to get the dynamics back; Dolby and others have some systems that can analyse sound data in real-time and try to re-create the dynamics. Hopefully the irony of this process is not lost on the reader and they can appreciate that it would be better if the dynamics had not been removed from the music in the first place!
Having said all of this, there are indeed several scenarios where DRC & limiting can enhance the listening experience:
1.) If the background acoustic noise levels are high (in a moving vehicle for example) DRC & limiting ensures you can hear all aspects of a music track without deafening yourself and your passengers during the loud parts.
2.) For late-night movie watching it can be used to ensure you can hear all the dialogue without waking your neighbours when the next bout of gunfire, explosions etc. hit the screen.
3.) It can be helpful when playing back music on smaller, less efficient loudspeakers such as those found inside smartphones, tablets and laptops.
Clearly the best place for DRC & limiting is on music playback devices themselves and *not* at the time the music/soundtrack is mastered. In this way, DRC & limiting can be applied when it is useful, and deactivated at other times for maximum audio quality.
Sign in to Reply
kdboyce
7/28/2011 4:12 PM EDT
Dr. HD,
The paragraphs make your post, and points, easier to read/understand. No apologies needed for them :)
The points you make in your first post are on the mark. Obviously if you can control the playback system, then you may be able to optimize the sound - from the mastering process down to the ears. This is why you will often find things like THX, and Dolby certified, etc. labels on consumer audio.
But in general, especially in portable devices, you don't have much control over the final hardware. Even to the extent that you might, the poor quality of the speakers will screw you up, as well as the adverse effects that can come from poor or limited acoustic design possible in such devices.
Sign in to Reply
Dr. HD
7/28/2011 5:23 PM EDT
Sure, when you're mastering music you don't have control over the quality of the playback system. But surely you're not suggesting that music or movie soundtracks should be mastered to the lowest common denominator? Mastering audio by DRCing and limiting it to death so that it sounds passable on poor quality playback hardware or in noisy listening environments inevitably makes said audio sound worse on even semi-decent equipment.
As a designer of high-quality audio power amplifiers the loudness war really depresses me. What is the point of trying to build power amps with 120 dB signal:noise ratio and 0.0001% THD if the music's dynamic range is only 6 dB or even worse? I hope that other manufacturers will follow Fairchild's lead. Having DRC and limiting, that the user can turn on and off when necessary, in the playback hardware itself would make it easier for the recording industry to de-escalate from the loudness war and hopefully begin again to master music with decent dynamics.
P.S. Really not sure what happened with the paragraphs thing: when I first posted my original comment, it appeared as one contiguous block of text and that's what I was apologising for. But when I later reloaded the page (after making my second post apologising for the poor formatting of the first), the paragraphs had magically re-appeared. It would be nice if we could edit and/or delete our comments because my second one's making me look like a bit of dolt!
Sign in to Reply
kdboyce
7/29/2011 2:40 AM EDT
No, I am not suggesting music be mastered to the lowest common denominator. I am very much for improving the playback equipment.
What I meant was that many of the products on the market now, and currently being designed, are being limited (no pun intended) more by the quality of the speakers and the inability to do really good acoustic design. Aside from size issues, many OEM's are not willing to put in really good speaker based audio since to do so may 'screw up' their nice industrial thin and light design. So they rely on headphone connections and very good SNR in that path.
National Semi, Wolfson, and TI have audio subsystems targeted at the portable device market, many of which incorporate one or more of the methods in the article
Sign in to Reply
Dwain
10/27/2011 10:21 AM EDT
Louder usually comes in a distorted form, when audio has less distortion, for a given power, it is often perceived as quieter.
Sign in to Reply