Method and apparatus for improving effective signal-to-noise ratio of analog to digital conversion for multi-band digital signal processing devices

Abstract

A method for improving the effective signal-to-noise ratio (“SNR”) of an analog to digital converter (“ADC”) for active loudspeakers uses the two available channels of a stereo ADC to separately process the low- and high-frequency components of an audio signal. Because the power spectral density of music approximates a pink noise spectrum, the high-frequency component of the signal has peak levels low enough to avoid exceeding the maximum ADC input level. The audio signal is analog high-pass filtered and the resulting high-frequency signal component is sent directly to a first ADC channel without attenuation. The remaining low-frequency component is attenuated and sent to a second ADC channel. The digital signals are processed, converted back to analog, amplified, and reproduced by loudspeaker drivers. Noise and distortion at low frequencies is less audible than higher frequencies, so the improved SNR at higher frequencies yields a significant practical improvement in audio fidelity.

Claims

1. A method for improving an effective signal-to-noise ratio of analog to digital audio signal conversion, the method comprising the steps of: receiving an input analog audio signal; high-pass filtering the input analog audio signal to produce a first high-frequency analog signal; converting the first high-frequency analog signal to a high-frequency digital signal; attenuating the input analog audio signal by approximately 15 dB to produce a first attenuated analog signal; converting the first attenuated analog signal to an attenuated digital signal; and applying digital signal processing to the high-frequency digital signal and/or the attenuated digital signal.

2. The method of claim 1, wherein the step of high-pass filtering the input analog audio signal comprises selecting a cutoff frequency of approximately 600 Hz.

3. The method of claim 1, wherein the step of high-pass filtering the input analog audio signal comprises selecting a cutoff frequency between 400-800 Hz.

4. The method of claim 1, wherein the step of applying digital signal processing comprises applying a digital high-pass filter to the high-frequency digital signal.

5. The method of claim 4, wherein the digital high-pass filter comprises a second-order high-pass filter.

6. The method of claim 4, wherein the digital high-pass filter comprises a plurality of cascaded high-pass filters.

7. The method of claim 1, wherein the step of applying digital signal processing comprises applying a digital low-pass filter to the attenuated digital signal.

8. The method of claim 7, wherein the digital low-pass filter comprises a second-order low-pass filter.

9. The method of claim 7, wherein the digital low-pass filter comprises a plurality of cascaded low-pass filters.

10. An analog to digital audio signal conversion system comprising: an analog audio input stage; an analog high-pass filter stage comprising a high-pass filter stage input connected to the analog audio input stage and a high-pass filter stage output; an analog attenuation stage comprising an attenuation stage input connected to the analog audio input stage and an attenuation stage output; a stereo analog to digital converter (ADC) comprising a first ADC channel input connected to the high-pass filter stage output, a second ADC channel input connected to the attenuation stage output, a first ADC channel output, and a second ADC channel output; a digital signal processor (DSP) comprising a first DSP channel input connected to the first ADC channel output, a second DSP channel input connected to the second ADC channel output, a first DSP channel output, and a second DSP channel output; and a stereo digital to analog converter (DAC) comprising a first DAC channel input connected to the first DSP channel output, a second DAC channel input connected to the second DSP channel output, a first DAC channel output, and a second DAC channel output.

11. The analog to digital audio signal conversion system of claim 10, wherein the analog high-pass filter stage further comprises a cutoff frequency of approximately 600 Hz.

12. The analog to digital audio signal conversion system of claim 10, wherein the analog high-pass filter stage further comprises a cutoff frequency between 400-800 Hz.

13. The analog to digital audio signal conversion system of claim 10, wherein the analog attenuation stage provides an attenuation of approximately 15 dB.

14. The analog to digital audio signal conversion system of claim 10, wherein the DSP further comprises a digital high-pass filter connected between the first DSP channel input and the first DSP channel output.

15. The analog to digital audio signal conversion system of claim 14, wherein the digital high-pass filter comprises a second-order high-pass filter.

16. The analog to digital audio signal conversion system of claim 14, wherein the digital high-pass filter comprises a plurality of cascaded high-pass filters.

17. The analog to digital audio signal conversion system of claim 10, wherein the DSP further comprises a digital low-pass filter connected between the second DSP channel input and the second DSP channel output.

18. The analog to digital audio signal conversion system of claim 17, wherein the digital low-pass filter comprises a second-order low-pass filter.

19. The analog to digital audio signal conversion system of claim 17, wherein the digital low-pass filter comprises a plurality of cascaded low-pass filters.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The present invention may be better understood, and its features made apparent to those skilled in the art by referencing the accompanying drawings.

(2) FIG. 1 is a schematic diagram of a prior art a two-way active loudspeaker with digital signal processing circuitry.

(3) FIG. 2 is a graph showing the power density versus frequency of a pink noise spectrum, as well as the attenuation versus frequency of a high-pass filter used in an embodiment of the present invention.

(4) FIG. 3 is a schematic diagram of a two-way active loudspeaker with digital signal processing circuitry having an improved effective signal-to-noise ratio, which is an embodiment of the present invention.

(5) The use of the same reference symbols in different drawings indicates similar or identical items.

DETAILED DESCRIPTION OF THE INVENTION

(6) A method for improving the effective signal-to-noise ratio of analog to digital and digital to analog conversion for active loudspeakers and other multi-band digital signal processing devices is presented. In one or more embodiments, the method of the present invention uses the two available channels of a stereo analog to digital converter device to separately process the low- and high-frequency components of the signal.

(7) The power spectral density of music approximates that of a pink noise (also known as 1/f noise) spectrum, i.e., one where the power density of the signal is inversely proportional to the frequency. Thus, for pink noise as well as for typical music, the peak signal power (and therefore peak signal level) at a particular frequency drops by 3 dB for every doubling of frequency, equivalent to a 30 dB difference between peak signal levels across the range of the audible spectrum from 20 Hz to 20 kHz.

(8) FIG. 2 is a graph 201 showing the power density versus frequency of a pink noise spectrum, as well as the attenuation versus frequency of a high-pass filter used in an embodiment of the present invention. The left vertical scale 202, right vertical scale 203, and horizontal scale 204 of graph 201 are all logarithmic. Left vertical scale 202 represents the power density in dB at a particular frequency, with the zero dB level normalized to the maximum power level of the entire signal. Right vertical scale 203 represents the signal attenuation in dB at a particular frequency of a high-pass filter having a cutoff frequency of approximately 600 Hz as used in an embodiment of the present invention.

(9) FIG. 2 shows that the power density for signal frequencies above the middle of the audio spectrum (i.e., around 600 Hz and higher) is approximately 15 dB lower than the power density for signal frequencies around 20 Hz. This is because the power density at a given frequency, represented by P.sub.f, is proportional to 1/f, so the difference in power density between 600 Hz and 20 Hz is

(10) $P_{600\mathrm{Hz}}-P_{20\mathrm{Hz}}=10{\log }_{10}\frac{P_{600\mathrm{Hz}}}{P_{20\mathrm{Hz}}}=10{\log }_{10}\frac{20\mathrm{Hz}}{600\mathrm{Hz}}\approx -14.8\mathrm{dB}.$
Therefore, the higher-frequency component of the audio signal (i.e., the audio frequencies above approximately 600 Hz) does not need to be attenuated by 15 dB before entering the ADC stage, because that higher-frequency component already has a peak signal level at least 15 dB lower than the peak signal level at 20 Hz.

(11) FIG. 3 is a schematic diagram of a two-way active loudspeaker with digital signal processing circuitry that exploits this property of the 1/f power density spectrum to improve the effective signal-to-noise ratio, which is an embodiment of the present invention. In the embodiment of FIG. 3, active loudspeaker 301 includes analog audio input stage 302, analog high-pass filter stage 303, analog attenuation stage 304, stereo analog to digital converter (“ADC”) 305, DSP 306, stereo digital to analog converter (“DAC”) 307, analog unity gain stage 308, analog signal booster stage 309, power amplifiers 310, and loudspeaker drivers 311.

(12) In the embodiment of FIG. 3, the audio input signal is fed to high-pass filter stage 303 along path 312 and separately to analog attenuation stage 304 along path 313 prior to any attenuation or analog to digital conversion. The input signal fed to high-pass filter stage 303 along path 312 is high-pass filtered in the analog domain by analog high-pass filter stage 303, then sent to first ADC channel 314 of ADC 305 to produce an unattenuated digital high-frequency component of the signal.

(13) In the embodiment of FIG. 3, analog high-pass filter stage 303 is an active second-order high-pass filter having unity gain that includes an operational amplifier and a resistive-capacitive network, with electronic component values chosen to place the cutoff frequency at approximately 600 Hz. In one or more alternative embodiments, analog high-pass filter stage 303 may have a gain of greater or less than unity. For example, analog high-pass filter stage 303 may attenuate the signal by a small amount, but by much less than the 15 dB of attenuation applied by analog attenuation stage 304. Alternatively, in one or more embodiments, analog high-pass filter stage 303 may be a passive high-pass filter or any other type of audio frequency filter. In one or more embodiments, the high-pass cutoff frequency required to avoid exceeding the input signal level limits of ADC 305 without signal attenuation is typically a value between 400-800 Hz, but analog high-pass filter stage 303 may have a higher or lower cutoff frequency as required to avoid exceeding the input limits of ADC 305.

(14) In the embodiment of FIG. 3, analog high-pass filter stage 303 performs a similar function to that of one of the second-order high-pass filters 114 shown in FIG. 1. For that reason, only one digital second-order high-pass filter 315 is included in DSP 306, with a unity filter stage 316 substituted for one of the second-order high-pass filters 114 shown in FIG. 1. In one or more alternative embodiments, unity filter stage 316 may be omitted, or additional or substitute first-order, second-order, or higher-order high-pass filter stages may be included in either or both of analog high-pass filter stage 303 or DSP 306 as required to achieve the desired crossover filtering function.

(15) In the embodiment of FIG. 3, the digital high-frequency component that is output from high-pass filter 315 is then converted back to an analog high-frequency signal component in a first channel of DAC 307. The analog high-frequency component is then passed through analog unity gain stage 308, is amplified by a first power amplifier 310, and is reproduced audibly by a first loudspeaker driver 311 (i.e., a tweeter). In one or more alternative embodiments, analog unity gain stage 308 may be omitted so that the output of the first channel of DAC 307 is routed directly to power amplifier 310.

(16) In one or more embodiments, the lack of attenuation of the high-frequency audio signal component allows DSP 306 to process that high-frequency component with a higher effective bit resolution. Furthermore, the high-frequency signal component does not need to be boosted after DAC 307, thereby avoiding the introduction of additional noise and distortion to the high-frequency component of the audio signal. In the embodiment of FIG. 3, the effective SNR of the high-frequency component is increased by 15 dB, or 5 bits of resolution.

(17) As demonstrated by FIG. 2, the low-frequency component of the audio signal must be attenuated so that the larger peak signal amplitude does not exceed the input range of ADC 305. In the embodiment of FIG. 3, the audio input signal that does not pass through analog high-pass filter 303 is attenuated by 15 dB in analog attenuation stage 304, then sent to second ADC channel 317 of ADC 305. In the embodiment of FIG. 3, the attenuated audio signal then passes through digital second-order low-pass filters 318 in DSP 306 to produce a digital low-frequency component. In one or more embodiments, additional or substitute first-order, second-order, or higher-order low-pass filter stages may be included in DSP 306 as required to achieve the desired crossover filtering function.

(18) In the embodiment of FIG. 3, the digital high-frequency component that is output from low-pass filters 318 is then converted back to an analog low-frequency signal component in a second channel of DAC 307. The analog low-frequency signal component is then boosted by analog signal booster stage 309, is amplified by a second power amplifier 310, and is reproduced audibly by a second loudspeaker driver 311 (i.e., a woofer). In one or more embodiments, although analog attenuation stage 304 and analog signal booster stage 309 introduce some additional noise and distortion to the low-frequency component of the audio signal, the noise and distortion is less audible than it would be for higher frequency audio content because human hearing is less sensitive to noise and distortion at low frequencies than at midrange and high frequencies. Thus, in the embodiment of FIG. 3, the improved SNR and effective bit depth in the midrange and high frequencies yields a significant practical improvement in overall loudspeaker performance and audio fidelity.

(19) In one or more embodiments, the audio signal may be split into more than two components. For example, in a three-way loudspeaker, the audio signal is split into low-, midrange-, and high-frequency components. In one or more embodiments, the low-frequency component is attenuated, digitized, digital low-pass filtered, converted back to analog, amplified, and routed to a woofer speaker driver as described above, but with a lowered low-pass filter cutoff of, for example, 300 Hz. Similarly, the high-frequency component is analog high-pass filtered, digitized, digital high-pass filtered, converted back to analog, amplified, and routed to a tweeter speaker driver as described above, but with a raised high-pass filter cutoff of, for example, 2000 Hz.

(20) Since the midrange frequencies require less attenuation than low frequencies, the midrange-frequency component may be analog band-pass filtered and attenuated by a smaller amount than the low-frequency component before entering the ADC. For example, in one or more embodiments, the analog midrange band-pass filter has a lower cutoff of 300 Hz and an upper cutoff of 2000 Hz, and the midrange frequency attenuation is only 3 dB. This is because the difference in power density between 300 Hz and 20 Hz is:

(21) $P_{300\mathrm{Hz}}-P_{20\mathrm{Hz}}=10{\log }_{10}\frac{P_{300\mathrm{Hz}}}{P_{20\mathrm{Hz}}}=10{\log }_{10}\frac{20\mathrm{Hz}}{300\mathrm{Hz}}\approx -11.7\mathrm{dB},$
thus requiring only approximately 3 dB attenuation to reach a signal level of −15 dB relative to 20 Hz. Alternatively, to reduce complexity and electronic component costs, the midrange-frequency component may only be analog high-pass filtered, for example with a cutoff of 300 Hz, with further band-pass filtering performed digitally within DSP 306. The midrange-frequency component is then digitized, digitally band-pass filtered, converted back to analog, amplified, and routed to a midrange speaker driver. Thus, in one or more embodiments, the SNR and effective bit depth may be optimized for multiple frequency bands, minimizing audible distortion even further than for the two-way speaker example.

(22) In one or more embodiments, the audio signal of path 313 may be analog low-pass filtered before attenuation to eliminate the energy content of the high-frequency component of the signal, thereby slightly reducing the attenuation required in analog attenuation stage 304 and the subsequent boost in analog signal booster stage 309. Although the high-frequency component of the signal adds only a small amount of additional energy to the signal, it is possible to save approximately 2 dB of headroom by filtering it out, thereby reducing the attenuation required and increasing the low-frequency resolution by 0.66 bits. In embodiments that require the highest audio fidelity, this improvement may be worth the added cost and complexity of the additional analog low-pass filters.

(23) In the embodiment of FIG. 3, analog audio input stage 302, analog high-pass filter stage 303, analog attenuation stage 304, and ADC 305 are shown with balanced signal inputs and outputs, which is commonly used in the line-level signal inputs and outputs of professional audio equipment to reduce the effect of external electromagnetic noise on the audio signal. In one or more alternative embodiments, analog audio input stage 302, analog high-pass filter stage 303, analog attenuation stage 304, and/or ADC 305 may instead use unbalanced signal inputs and/or outputs (i.e., a single-ended signal wire and a ground, as is commonly used in consumer-grade audio equipment) for all or part of the pre-DSP signal path. Similarly, in the embodiment of FIG. 3, gain stages 308 and 309 and power amplifiers 310 are shown with unbalanced signal inputs and outputs, but may instead use balanced signal inputs and/or outputs for all or part of the post-DSP signal path in one or more alternative embodiments.

(24) Thus, a method for improving the effective signal-to-noise ratio of analog to digital and digital to analog conversion for active loudspeakers and other multi-band digital signal processing devices by using the two available channels of a stereo analog to digital converter device to separately process the low and high-frequency components of the signal is described. Although the present invention has been described with respect to certain specific embodiments, it will be clear to those skilled in the art that the inventive features of the present invention are applicable to other embodiments as well, all of which are intended to fall within the scope of the present invention. For example, the cutoff frequencies of the low-pass, band-pass, and/or high-pass filters may be adjusted to suit the frequency response range of each speaker driver. Similarly, the amount of pre-ADC attenuation and/or post-DAC boost may be adjusted according to the maximum peak-to-peak input signal level and the maximum allowable signal level for the ADC. Additionally, the method may be used to improve the effective signal-to-noise ratio in any application that uses multi-band digital signal processing, such as audio compressors, audio effects processors, audio and/or video recording devices, sound reinforcement or public address systems, or speech recognition, among others.