yamamoto2002 Posted February 19, 2022 Share Posted February 19, 2022 It is a bit difficult to hear the sound difference of digital filtering methods (sharp roll-off, slow roll-off, NOS), because the difference happens on very high frequency range. There is a method to enhance the sound difference to hear by ears clearly: shift sound pitch by 15vb (2 octaves) to move it to hearing range. Equipment used RME ADI-2 Pro as DAC, XLR analog line-out, it has several different digital filter settings. RME Babyface Pro as ADC, XLR analog line-in 1. Sound file used It is my original recorded sound of a bar chime, using AKG C414 XLS, Babyface Pro, 24bit 96kHz 1ch PCM. barchime2496.flac Original file down-sampled to 24bit 44.1kHz using sox "sox barchime2496.wav barchime2444_sox997.wav rate -b 99.7 44100" barchime2444_sox997.flac Experiment Procedure Play the sound "barchime2444_sox997" with a DAC with various digital filter settings. Record the analog line-level sound with ADC to create 192kHz 24bit PCM. Open recorded wav file with your favorite binary editor and edit the WAV header part "SampleRate" from 192000 (0x2ee00) to 48000 (0xbb80) and "ByteRate" from 576000 (0x8ca00) to 144000 (0x23280) and save it as different file. N.B. WAV header integer values are stored as little-endian byte order. (Fig.1 and Fig.2) Play the resulted files and compare the sound by ears. Sound files for listening comparison Played barchime2444 with RME ADI-2 Pro "Sharp digital filter", record it as 192kHz 24bit PCM then shift the pitch by 15vb. 2444adi2Sharp_24192babyface_15vb.flac Played barchime2444 with RME ADI-2 Pro "Slow digital filter", record it as 192kHz 24bit PCM then shift the pitch by 15vb. 2444adi2Slow_24192babyface_15vb.flac Played barchime2444 with RME ADI-2 Pro "NOS digital filter", record it as 192kHz 24bit PCM then shift the pitch by 15vb. 2444adi2NOS_24192babyface_15vb.flac Original file, pitch shifted by 15vb. barchime2496_15vb.flac Fig.1 Find "SampleRate" and "ByteRate" of WAV header. Fig.2 Divide the values by 4 to create 15vb files for listening comparison. Fig.3 "barchime2496.flac" original recorded sound spectrogram. 32kHz or above frequency exhibits the distortion of the microphone, therefore it is attenuated using Adobe Audition. Fig.4 "barchime2496_15vb.flac", Playback time becomes 4x longer and all signal frequency becomes 4x lower. Fig.5 "barchime2444_sox997.flac", original file downsampled to 24bit 44.1kHz PCM. Notice 22.5kHz or above signal is deleted by downsampling. Fig.6 Sharp roll-off digital filter. This is a text-book digital filter. The information stored on the file "barchime2444_sox997.flac" is outputted faithfully. Vertical axis range is now becomes 0 to 96kHz because recording sample rate is 192kHz. Fig.7 NOS (zero order hold) digital filter. Notice the aliasing noise above 22.05kHz, the frequency range 22.05kHz to 44.1kHz is vertically flipped image of 0 to 22.05kHz and 44.1kHz to 66.15kHz is vertically flipped image of 22.05kHz to 44.1kHz and so on. those signal is not found on playback file nor the original 96kHz recorded file, so it is unwanted noise IMO. The aliasing noise sound can be very annoying because its frequency is not related to the original signal, but it is not awful very much in this example sound. Fig.7 Slow roll-off digital filter. Notice the aliasing noise above 22.05kHz. The aliasing noise is visible (above 22.05kHz), its amount is smaller than NOS. Notice the aliasing noise signal is not related to the original signal of harmonics. Josh Mound 1 Sunday programmer since 1985 Developer of PlayPcmWin Link to comment
yamamoto2002 Posted February 19, 2022 Author Share Posted February 19, 2022 I found 15vb is a wrong notation, 15mb (abbreviation of quindicesima bassa) is correct. Sunday programmer since 1985 Developer of PlayPcmWin Link to comment
Guest Posted February 20, 2022 Share Posted February 20, 2022 Analyzing Sound vs Listening to Music Yes there is a line between the two, sometimes we drift from one side to the other. Link to comment
bobbmd Posted April 26, 2022 Share Posted April 26, 2022 @yamamoto2002—It all depends on one’s YEARS—get it ‘ears’/‘years’ ? Link to comment
pkane2001 Posted April 26, 2022 Share Posted April 26, 2022 On 2/19/2022 at 5:35 AM, yamamoto2002 said: It is a bit difficult to hear the sound difference of digital filtering methods (sharp roll-off, slow roll-off, NOS), because the difference happens on very high frequency range. There is a method to enhance the sound difference to hear by ears clearly: shift sound pitch by 15vb (2 octaves) to move it to hearing range. Equipment used RME ADI-2 Pro as DAC, XLR analog line-out, it has several different digital filter settings. RME Babyface Pro as ADC, XLR analog line-in 1. Sound file used It is my original recorded sound of a bar chime, using AKG C414 XLS, Babyface Pro, 24bit 96kHz 1ch PCM. barchime2496.flac 1.09 MB · 14 downloads Original file down-sampled to 24bit 44.1kHz using sox "sox barchime2496.wav barchime2444_sox997.wav rate -b 99.7 44100" barchime2444_sox997.flac 625.17 kB · 17 downloads Experiment Procedure Play the sound "barchime2444_sox997" with a DAC with various digital filter settings. Record the analog line-level sound with ADC to create 192kHz 24bit PCM. Open recorded wav file with your favorite binary editor and edit the WAV header part "SampleRate" from 192000 (0x2ee00) to 48000 (0xbb80) and "ByteRate" from 576000 (0x8ca00) to 144000 (0x23280) and save it as different file. N.B. WAV header integer values are stored as little-endian byte order. (Fig.1 and Fig.2) Play the resulted files and compare the sound by ears. Sound files for listening comparison Played barchime2444 with RME ADI-2 Pro "Sharp digital filter", record it as 192kHz 24bit PCM then shift the pitch by 15vb. 2444adi2Sharp_24192babyface_15vb.flac 2.09 MB · 14 downloads Played barchime2444 with RME ADI-2 Pro "Slow digital filter", record it as 192kHz 24bit PCM then shift the pitch by 15vb. 2444adi2Slow_24192babyface_15vb.flac 2.15 MB · 13 downloads Played barchime2444 with RME ADI-2 Pro "NOS digital filter", record it as 192kHz 24bit PCM then shift the pitch by 15vb. 2444adi2NOS_24192babyface_15vb.flac 2.66 MB · 13 downloads Original file, pitch shifted by 15vb. barchime2496_15vb.flac 1.09 MB · 16 downloads Fig.1 Find "SampleRate" and "ByteRate" of WAV header. Fig.2 Divide the values by 4 to create 15vb files for listening comparison. Fig.3 "barchime2496.flac" original recorded sound spectrogram. 32kHz or above frequency exhibits the distortion of the microphone, therefore it is attenuated using Adobe Audition. Fig.4 "barchime2496_15vb.flac", Playback time becomes 4x longer and all signal frequency becomes 4x lower. Fig.5 "barchime2444_sox997.flac", original file downsampled to 24bit 44.1kHz PCM. Notice 22.5kHz or above signal is deleted by downsampling. Fig.6 Sharp roll-off digital filter. This is a text-book digital filter. The information stored on the file "barchime2444_sox997.flac" is outputted faithfully. Vertical axis range is now becomes 0 to 96kHz because recording sample rate is 192kHz. Fig.7 NOS (zero order hold) digital filter. Notice the aliasing noise above 22.05kHz, the frequency range 22.05kHz to 44.1kHz is vertically flipped image of 0 to 22.05kHz and 44.1kHz to 66.15kHz is vertically flipped image of 22.05kHz to 44.1kHz and so on. those signal is not found on playback file nor the original 96kHz recorded file, so it is unwanted noise IMO. The aliasing noise sound can be very annoying because its frequency is not related to the original signal, but it is not awful very much in this example sound. Fig.7 Slow roll-off digital filter. Notice the aliasing noise above 22.05kHz. The aliasing noise is visible (above 22.05kHz), its amount is smaller than NOS. Notice the aliasing noise signal is not related to the original signal of harmonics. NOS absence of digital filter is often an audible problem. It causes potentially high-in-level images that can intermodulate into the audible range, plus frequency response fall-off, also within the audible range. Some DACs (like RME ADI-2) support a NOS-like filter but provide a frequency fall-off compensation. But that still doesn't address the images or their possible IMD interaction. I do like your method of adjusting the sample rate! Easy enough to do. With DeltaWave app, I implemented a frequency-domain adjustment that allows one to shift frequencies from higher to lower range by an arbitrary amount and then listen. It is interesting to hear what type of sounds/noise/patterns are contained above the audible range. Josh Mound 1 -Paul DeltaWave, DISTORT, Earful, PKHarmonic, new: Multitone Analyzer Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now