Bass Extension Comparison: Waves MaxxBass and SRS TruBass TM

Transcription

1 Bass Extension Comparison: Waves MaxxBass and SRS TruBass TM Meir Shashoua Chief Technical Officer Waves, Tel Aviv, Israel Paul Bundschuh Vice President of Marketing Waves, Austin, Texas USA 1.0 Introduction Waves patented MaxxBass psycho-acoustic bass extension technology is a valuable feature in a wide variety of consumer audio electronics to compensate for electroacoustic bass limitations due to cost, size and power consumption. MaxxBass exploits the Phenomenon of the Missing Fundamental to generate a perceived bass extension of up to 1 and 1/2 octaves. MaxxBass has been used for years on many of the world s most popular music to improve bass lines and mix compatibility. Waves has recently began shipping a low cost ASIC, MX3000AS, making MaxxBass attractive for cost sensitive consumer electronics. 1.1 Tests Performed Several tests and comparisons were made between two bass extension technologies, MaxxBass from Waves and TruBass TM from SRS Labs. This paper provides detailed test results on frequency sweep and pure sine wave tests. Tests were performed identically on both systems using commercially available products. Professional test equipment was used for all measurements. The tests measure audio system response that would be expected when using these bass extension technologies with small speakers with a 100 Hz physical cutoff frequency typical of televisions and multimedia speakers. The frequency sweep tests used a slow sweep from 20 Hz to 20 khz with a 1Vpp input. These tests show system linearity and any changes in level. The frequency response for pure sine waves was measured to show the harmonics generation at several frequencies and levels. Sine waves at 50, 70 and 100 Hz was measured to demonstrate how the harmonics generation varies with frequency. Several input levels were used at each frequency to demonstrate if the harmonics are generated consistently across the wide dynamic range of music. The output plots from all the tests are included in this paper. The authors have also provided analysis for both technologies as well as some general conclusions on the superiority of MaxxBass in providing more consistent results across frequencies and levels and without a substantial impact to the system s signal-to-noise ratio (SNR). Bass Extension Comparison 1 February 2002

2 2.0 MaxxBass Pro102 Test Set-Up Waves MaxxBass Pro102 was the MaxxBass test system used. This product uses the Waves MX3000AS MaxxBass ASIC. The same MaxxBass Pro102 settings were made for all tests, except when Bypass Mode was used. All tests were made with no input attenuation, which corresponds to clock position 5 on the Input Signal knob. The tests were made at maximum system intensity, which corresponds to clock position 5 on the Intensity knob. The MaxxBass frequency was set to 100 Hz, which corresponds to a clock position of 1 on the Frequency knob. The Bypass Mode is enabled by pressing the bypass button to the left of the power LED. (When the system is in Bypass Mode the Bypass LED is illuminated.) The test settings on the MaxxBass Pro102 are shown in Figure 1. Figure 1. MaxxBass Pro102 Test Settings 2.1 SRS WowThing TM Test Set-Up SRS offers a commercial product called WowThing that provides TruBass functionality. It utilizes the AP9883 device from SRS semiconductor subsidiary. The same WowThings settings were made for all tests, except when Bypass Mode was used. The WOW knob was turned completely to the left to minimize the WOW functions, which corresponds to clock position 7. The TruBass function was maximized by turning the TruBass knob completely to the right, which corresponds to clock position 5. Volume was originally set to full scale, but it was lower to allow listening tests without clipping in the speaker to due excess signal gain. The volume was lowered to 14dB for all tests in order to compensate for TrueBass +20dB low frequency gain and allow listening without clipping in the speaker system. The Bypass Mode is enabled by moving the Bypass/Wow switch to the left. The test settings on the WowThing are shown in Figure 2. Bass Extension Comparison 2 February 2002

3 Figure 2. WOWThing Test Settings WOW THING TM OFF/ON BYPASS/WOW WOW TruBass VOLUME Bass Extension Comparison 3 February 2002

4 3.0 Frequency Sweep System Calibration The frequency sweep tests are to measure the frequency response for linearity as well changes in level. Frequency sweep tests were made to test for system calibration in addition for both Bypass Mode and ON modes for both technologies. In all cases a 1Vpp input signal was used. In the calibration test neither the MaxxBass Pro102 nor WowThing box was physically connected into the audio system. Figure 3 shows the results when the frequency generator output is connected directly to the input of the measurement system. The test and measurement system is practically flat from 20 Hz to 20 khz. The 1Vpp maps to the level of -13dB. The -13dB level represents no system attenuation in all the other plots presented in this paper. Figure 3. Calibration for Test System Linearity and Level Bass Extension Comparison 4 February 2002

5 4.0 MaxxBass Frequency Sweep Tests 4.1 MaxxBass Bypass Mode Frequency Response From the MaxxBass Bypass Mode plot (Figure 4) it is apparent that the MaxxBass Bypass Mode has a flat response with a 1:1 gain throughout the entire frequency spectrum. The 1Vpp input corresponds to a level of -13dB level, the same as during the test calibration so no system attenuation occurs during Bypass Mode. The MaxxBass Bypass Mode frequency response practically matches the Test Calibration (Figure 3), so that during Bypass Mode of the MX3000AS the audio system performance is equivalent to not physically including MX3000AS in the audio circuit. Figure 4. MaxxBass Bypass Mode with 1Vpp 4.2 MaxxBass Bypass Mode Frequency Response From the MaxxBass ON measurement (Figure 5) several important characteristics of MaxxBass processing can be seen. 1) Above 200 Hz the frequency response is completely flat. MaxxBass makes no effect on these frequencies leaving the majority of the audio signal unaltered. 2) At 100 Hz (with MaxxBass Fc=100Hz) a modest boost of +6 db is added. The +6 db boost near the speaker cutoff frequency compensates for the first -6 db drop in frequency response of the speaker, just below the speaker cutoff. For the first -6 db of speaker drop off MaxxBass can compensate by simply boosting the fundamental, since the speaker only has a modest loss of efficiency very near the speaker cutoff. Bass Extension Comparison 5 February 2002

6 A boost of +6 db is modest since it lies inside the tolerance of speakers and amplifiers, while providing good results in the immediate frequency region of the speaker s cutoff frequency. Although MaxxBass adds +6 db of gain at 100 Hz and harmonics at higher frequencies the measured increase in speaker excursion is only +3-4 db, since much of the +6 db boost occurs at or below the physical cutoff where the speaker efficiency is already starting to drop. The +3-4 db increase in speaker excursion is well inside speaker margin requirements of most systems. Additionally attempting to boost beyond beyond this modest level can create distortions in the speaker as later discussed. 3) Below 100 Hz a -18 db/octave high pass filter function is seen. The -18 db/octave High Pass Filter below Fc is provided in order to remove excess low frequency energy from the speaker, as the speaker is incapable of reproducing them anyway. The frequencies below Fc only stress the speaker and create Inter-Modulation- Distortions for higher frequencies, so it is better to remove these frequencies completely before amplification and obtain the additional system benefit of a reduction system in system power consumption. It is clear that MaxxBass affects only frequencies from about Fc (100Hz in this case) and below. As a bass enhancement technology MaxxBass only affects the low frequency audio signals near or below the physical speaker cutoff and leaves the majority of the audio signals unaltered. Figure 5. MaxxBass ON with 1Vpp Bass Extension Comparison 6 February 2002

7 5.0 TruBass Frequency Sweep Tests 5.1 TruBass Bypass Mode Frequency Response From the TruBass Bypass Mode test (Figure 6) it is clear that the Bypass Mode frequency response is flat. Attenuation of -14 db was needed to provide the necessary system headroom for when the TruBass function is ON; at least this much headroom will be needed for the audio system to cope with the +20 db low-frequency boost inserted by TrueBass. This costs the audio system -14 db of system signal-to-noise (SNR) seriously degrading the audio system even when the TruBass function is off. Figure 6. TruBass Bypass Mode with 1Vpp 5.2 TruBass ON Frequency Response with 1Vpp When TruBass is ON with 1Vpp input (Figure 7) it can be seen that the frequency response is no longer close to flat. 1) A huge +20dB bass boost has been added below 250Hz. 2) A more modest +6B high frequency boost has been added above 1 khz. 3) A relative sharp notch filter has been added at about 570 Hz. Bass Extension Comparison 7 February 2002

8 Figure 7. TruBass ON with 1Vpp 5.3 TruBass ON Frequency Response with 250mVpp When the input signal is lowered to 250mVpp (Figure 8) the entire curve (of Figure 7) is simply shifted down by -12 db, which is the exact ratio between 1Vpp and 250mVpp. It is clear from Figures 6-8 that TruBass employs a fixed boost filter for the frequencies below 250 Hz. If a dynamic boost was utilized then the shape of the curve would vary based on input level. Since only a level shift of 12 db is seen, it can be concluded that TruBass uses a fixed boost filter. To confirm linearity of the bass boost, a further test was implemented on Audio Precision test equipment. The input signal was a 60 Hz tone, which is the peak of the TruBass boost curve, and the amplitude was varied between 10mVpp to 1Vpp. Across this range of levels the output was linearly related to the input. Bass Extension Comparison 8 February 2002

9 Figure 8. TruBass ON with 250mVpp 5.4 System Implications for TruBass +20 db bass boost TruBass uses a fixed +20 db bass boost method to achieve of its perceived bass improvement. It is a well known psycho-acoustic principle that increasing the volume improves the users general perception of sound; however using such a large bass boost has several well known negative audio system implications. To practically utilize TruBass +20 db bass gain, the audio system engineer must attenuate the overall audio level by at least -14 db to create the necessary audio headroom. (This also assumes the audio system already has 6 db of headroom margin that can also be used.) The major drawback of this approach is that the system SNR is also reduced by at least 14 db seriously degrading the quality of the audio whether TruBass is ON or in Bypass Mode. A strong bass boost function such as TruBass with high input level signal can result in distorted sound and possibly overloading the amplifier and the speaker. There is an increased probability of burning out the speaker and/or the amplifier unless more powerful amplifiers and more expensive speakers with larger excursions are utilized. This can substantially increase the system cost of implementing a TruBass system. Bass boost systems also do not improve the perceived linearity of frequency response, which is the normally the audio system designers goal, since they create a substantial gain only near the speaker cutoff frequency. This is also clear apparent in listening tests with TruBass. Bass Extension Comparison 9 February 2002

10 6.0 MaxxBass Sine Wave Tests Pure sine wave inputs are used to measure harmonics generated by the MaxxBass functions. Sine waves at 100 Hz, 70 Hz and 50 Hz were used as inputs at levels of 1.4Vpp, 1.0Vpp, 500mVpp, 250mVpp, and 100mVpp. 6.1 MaxxBass 100 Hz Sine Wave Tests Figures show the response of MaxxBass to 100 Hz pure sine. For this test the MaxxBass Frequency (Fc) is set to 100 Hz, so the sine wave is at the upper edge of the frequency range for which MaxxBass processing is applied. Additionally the physical speaker cutoff is again assumed also to be about 100 Hz, so most of the energy output is still at the 100 Hz fundamental frequency and relatively little energy is in the harmonics. These plots illustrate that the 100 Hz fundamental frequency does not scale linearly with the input level due to dynamic range compression applied as part of the MaxxBass process. The level of harmonics is in a consistent ratio to the level of the fundamental at every input level. This is particularly noticeable and important on the first and second harmonics, at 200 Hz and 300 Hz. The level of harmonics is carefully computed, in order to achieve natural sounding dynamics for the harmonics signal, and match that of the original sound. Figure 10. MaxxBass 100 Hz, 1.4Vpp Bass Extension Comparison 10 February 2002

11 Figure 11. MaxxBass 100 Hz, 1.0Vpp Figure 12. MaxxBass 100 Hz, 500mVpp Bass Extension Comparison 11 February 2002

12 Figure 13. MaxxBass 100 Hz, 250mVpp Figure 14. MaxxBass 100 Hz, 100mVpp Bass Extension Comparison 12 February 2002

13 6.3 MaxxBass 70 Hz Sine Wave Tests Figures show the response of MaxxBass to 70Hz pure sine. As 70Hz is lower compared to Fc (100Hz), it can be seen that more energy at the output shifted towards the harmonics. As with the case at 100 Hz, the fundamental frequency at 70 Hz does not scale linearly with the input level due to dynamic range compression applied as part of the MaxxBass process. The level of harmonics is in a consistent ratio to the level of the fundamental at every input level. This is even more apparent at 70 Hz where the fundamental and all the harmonics have shifted down with input. The level of harmonics is carefully computed, in order to achieve natural sounding dynamics for the harmonics signal, and match that of the original sound. Figure 15. MaxxBass 70 Hz, 1.4Vpp Bass Extension Comparison 13 February 2002

14 Figure 16. MaxxBass 70 Hz, 1.0Vpp Figure 17. MaxxBass 70 Hz, 500mVpp Bass Extension Comparison 14 February 2002

15 Figure 18. MaxxBass 70 Hz, 250mVpp Figure 19. MaxxBass 70 Hz, 100mVpp Bass Extension Comparison 15 February 2002

16 6.5 MaxxBass 50 Hz Sine Wave Tests Figures show the response of MaxxBass to 50Hz pure sine wave. 50Hz is substantially lower than Fc (100Hz), it can be seen that even more energy is shifted into the harmonics. At 50 Hz the harmonics actually have more energy than the fundamental. As with the case at 70 and 100 Hz, the fundamental frequency at 50 Hz does not scale linearly with the input level due to dynamic range compression applied as part of the MaxxBass process. The level of harmonics is in a consistent ratio to the level of the fundamental at every input level and the entire plot shifts downward with input level. The level of harmonics is carefully computed, in order to achieve natural sounding dynamics for the harmonics signal, and match that of the original sound. Figure 20. MaxxBass 50 Hz, 1.4Vpp Bass Extension Comparison 16 February 2002

17 Figure 21. MaxxBass 50 Hz, 1.0Vpp Figure 22. MaxxBass 50 Hz, 500mVpp Bass Extension Comparison 17 February 2002

18 Figure 23. MaxxBass 50 Hz, 250mVpp Figure 24. MaxxBass 50 Hz, 100mVpp Bass Extension Comparison 18 February 2002

19 7.0 TruBass Sine Wave Tests Pure sine wave inputs are used to measure harmonics generated by the TruBass functions. Sine waves at 100 Hz, 70 Hz and 50 Hz were used as inputs at various input levels to evaluate the consistency of the harmonics generation. 7.1 TruBass 100 Hz Sine Wave Tests Figures show TruBass response to a 100 Hz pure sine wave input at several levels. The following characteristics can be seen. 1. The fundamental 100Hz tone always has most of the energy. The fundamental 100 Hz shows a gain of more than +20dB from the fixed bass boost functionality described previously. 2. Harmonics are generated, but in an inconsistent manner to either the input or the fundamental. At 700mVpp levels and below no harmonics are generated. 3. At 1.4Vpp the level of very high-frequency harmonics is too high. Too much high frequency energy (up to 1 khz) is added back into the audio much of which does not occur at specific harmonic frequencies. The result is a BUZZ sound, rather than a deep bass sound. 4. At 1.0Vpp level inputs there is a reasonable amount of good harmonics to satisfy the missing fundamental effect; however this occurs only in a very limited range of about 3 db. The level of harmonics does not scale continuously with input level and the range of acceptable harmonics generation is so small (only 3 db), that practically speaking with normal music there will never be a good missing fundamental effect. Above the small +3 db input level range, too much high harmonics are generated. This behavior is similar to the effect of an analog clipper, where the 3dB range is the transition between no clipping (and no harmonics) and severe clipping (and too much harmonics). Across the normal dynamic range of music, the TruBass bass improvement is practically only relying on a simple fixed bass boost, which can be achieved without the expense of a custom circuit. Bass Extension Comparison 19 February 2002

20 Figure 25. TruBass 100 Hz, 1.4Vpp Figure 26. TruBass 100 Hz, 1.0Vpp Bass Extension Comparison 20 February 2002

21 Figure 27. TruBass 100 Hz, 700mVpp Figure 28. TruBass 100 Hz, 500mVpp Bass Extension Comparison 21 February 2002

22 Figure 29. TruBass 100 Hz, 250mVpp Figure 30. TruBass 100 Hz, 100mVpp Bass Extension Comparison 22 February 2002

23 7.2 TruBass 70 Hz Sine Wave Tests Figures show TruBass response to a 70 Hz pure sine wave input with several amplitudes. At 70 Hz the sine wave tests show the same issues with TruBass as discussed for 100 Hz. Figure 31. TruBass 70 Hz, 1.4Vpp Figure 32. TruBass 70 Hz, 1.0Vpp Bass Extension Comparison 23 February 2002

24 Figure 33. TruBass 70 Hz, 700mVpp Figure 34. TruBass 70 Hz, 500mVpp Bass Extension Comparison 24 February 2002

25 Figure 35. TruBass 70 Hz, 250mVpp Figure 36. TruBass 70 Hz, 100mVpp Bass Extension Comparison 25 February 2002

26 7.3 TruBass 50 Hz Sine Wave Tests Figures show TruBass response to a 50 Hz pure sine wave input with several amplitudes. At 50 Hz the sine wave tests show the similar issues at 70 and 100 Hz, although the range of good harmonics has shifted down to mVpp. Above this range at 1Vpp too much high-harmonics are generated, and below this, at 250mVpp no harmonics are generated. Figure 37. TruBass 50 Hz, 1.4Vpp Figure 38. TruBass 50 Hz, 1.0Vpp Bass Extension Comparison 26 February 2002

27 Figure 39. TruBass 50 Hz, 700mVpp Figure 40. TruBass 50 Hz, 500mVpp Bass Extension Comparison 27 February 2002

28 Figure 41. TruBass 50 Hz, 250mVpp Bass Extension Comparison 28 February 2002

29 8.0 Conclusions on Bass Extension Comparisons 8.1 Bass Frequency Constraints and Goals The goal of the audio system designer is to provide linear response in audio electronics, amplifiers and speakers systems, while meeting product constraints on cost, size and power consumption. As the product constraints tighten the audio system engineer is under pressure to raise the speaker cutoff frequency, which reduces the bass frequency response of the audio system. This is due to an electro-acoustic limitation in which the speakers frequency response is linear above the speaker cutoff frequency and attenuates -12 to -24 db/octave below this frequency. The goal of any bass extension method should be to extend the perceived bass frequency linearity below the speaker cutoff frequency, while having minimal impact on the both the linearity and maximum level of the audio system. 8.2 Waves MaxxBass Results MaxxBass has been demonstrated to increase the bass frequency response up to 1 and ½ octaves below the speaker cutoff frequency. This is accomplished without changing the frequency response linearity above the speaker cutoff frequency and without requiring additional system headroom margin required by large bass boosts. A listener perceived linear extension of bass frequency can be achieved through use of a patented algorithm that uses the Phenomenon of the Missing Fundamental. The perceived natural quality of this algorithm is apparent through listening tests and the MaxxBass algorithm has been used for years in the production of the world s most popular music. Consumer electronic manufacturers can now benefit from the MaxxBass technology with a new low cost MX3000AS ASIC. 8.3 SRS TruBass Results TruBass utilizes a fixed bass boost of +20 db to generate its dominant bass effect across the dynamic range of music. To provide adequate system headroom to utilize this boost, the audio system must sacrifice at least -14 db of system SNR seriously degrading the quality of audio whether TruBass is ON or in Bypass Mode. This large boost does not improve perceived bass linearity, but merely increases the volume of the bass near the cutoff frequency. This creates unnatural sounding bass particularly at high levels, which is apparent in critical audio listening tests. TruBass frequency response also show a high frequency gain and a notch filter at approximately 570 Hz further reducing the goal of audio system linearity. TruBass implements some harmonics generation, but the effective input level range is only +3 db making it practically useless for the large dynamic range of audio used in music, movie soundtrack or games. Bass Extension Comparison 29 February 2002