Excellent point. Fortunately, that's something you can do manually in Goldwave.
Well that puts my mind at ease
For what its worth I like Goldwave and used to use it a lot, but I wouldn't use it for a critical ABX comparison of different sample rates when the resampling algorithm is fancy sounding but undisclosed
Even if it was a well meaning algorithm it could still have bugs, for example a bug that causes the wrong amplitude of dither to be applied during resampling could easily go unnoticed without a strong suite of test signals...
To confirm correctness we would need something with source code available and compilable.
Nope, I don't have a specific suggestion at the moment, just pointing out that I'm wary of using a "black box" to do the resampling for us when there are so many different possible algorithms and we really don't know what its doing inside.If I heard a difference, I'd agree.
If I was trying to set up some kind of scientifically valid and verifiable listening experiment I would definitely use something where source code was available, although of course you need someone with both strong programming skills and excellent signal processing theory to be able to look at the code to verify its correctness...(in other words not me
)
In case there is no consensus for resampling I mostly use r8brain PRO by Voxengo. There will probably be some faults in that one too, and it does only "flat" dither, but I am mostly very happy with the results. It is also ancient in computer years. My version is from 2006. No relation to the product or its maker. I cant remember it being cheap but hey I surely am so that does not mean a lot. There does seem to be a demo version. Comparisons can be found at SRC Comparisons
From the help file:
r8brain PRO is a professional sample rate converter designed to deliver an unprecedented sample rate conversion (SRC) quality. Unlike many existing SRC algorithms available on the market, r8brain PRO implements sample rate conversion processing in its full: interpolation and decimation steps without exploiting any kind of simplifications; the signal is first resampled to a least common multiple sample rate which makes conversion perfect. In the core of the SRC algorithm, we use the convolution methods of our Pristine Space convolution processor which is known for its highly precise convolution processing. This gives us high sample rate conversion quality in combination with comparably small processing times: sample rate conversion without compromises!
Like many existing SRC programs, r8brain PRO offers you a linear-phase conversion mode. But more importantly, you also have an option of using the minimum-phase conversion mode, which finally brings SRC with true analog qualities to affordable digital audio workstations: in this mode, r8brain PRO works like an ideal digital-to-analog converter followed by an analog-to-digital converter to resample the audio. This eliminates pre-ringing associated with linear-phase designs, while introducing only a minimal amount of phase coloration.
r8brain PRO can read mono, stereo and multi-channel files in both WAV and AIFF file formats, creating 16-, 24- and 32-bit mono, stereo and multi-channel WAV files in fixed and floating point formats. EBU BWF (broadcasting) extensions, extensible wave format, sample loops and textual data residing inside the file are also supported. For the sake of convenience, r8brain PRO allows you to perform batch conversions in a convenient manner.
r8brain PRO's bit-depth conversion is limited to flat dithering. We have decided not to implement noise-shaping dithering because pro audio production software available on the market usually offers the user noise-shaping dithering of some kind already. We also based our decision on the fact that the sample rate conversion process often adjusts peak structure of the original program material, thus, in many cases, making a subsequent peak-limiting a necessity. To prevent output audio from clipping we have implemented a level normalization feature.
The demo version of r8brain PRO allows you to evaluate all features of the software, but its conversion is limited to the first minute of an audio file. The demo version does not support batch conversions. You may register to obtain either a Full or Light license of r8brain PRO. The Light version offers you a linear-phase conversion mode and up to 48 kHz output sample rates only while supporting all input sample rates. The Full version offers you all features of the software.
r8brain PRO features:
# Reads 8-, 16-, 24-, 32-bit PCM and IEEE files
# Reads both WAV and AIFF files
# Writes 16-, 24-, 32-bit PCM and IEEE WAV files
# Multi-channel file support
# Extensible wave format support
# EBU BWF extensions support
# Linear- and minimum-phase modes
# Ultra-steep conversion mode
# Batch support
# Automatic normalization
# Supports all standard sample rates
From the help file:
r8brain PRO is a professional sample rate converter designed to deliver an unprecedented sample rate conversion (SRC) quality. Unlike many existing SRC algorithms available on the market, r8brain PRO implements sample rate conversion processing in its full: interpolation and decimation steps without exploiting any kind of simplifications; the signal is first resampled to a least common multiple sample rate which makes conversion perfect. In the core of the SRC algorithm, we use the convolution methods of our Pristine Space convolution processor which is known for its highly precise convolution processing. This gives us high sample rate conversion quality in combination with comparably small processing times: sample rate conversion without compromises!
Like many existing SRC programs, r8brain PRO offers you a linear-phase conversion mode. But more importantly, you also have an option of using the minimum-phase conversion mode, which finally brings SRC with true analog qualities to affordable digital audio workstations: in this mode, r8brain PRO works like an ideal digital-to-analog converter followed by an analog-to-digital converter to resample the audio. This eliminates pre-ringing associated with linear-phase designs, while introducing only a minimal amount of phase coloration.
r8brain PRO can read mono, stereo and multi-channel files in both WAV and AIFF file formats, creating 16-, 24- and 32-bit mono, stereo and multi-channel WAV files in fixed and floating point formats. EBU BWF (broadcasting) extensions, extensible wave format, sample loops and textual data residing inside the file are also supported. For the sake of convenience, r8brain PRO allows you to perform batch conversions in a convenient manner.
r8brain PRO's bit-depth conversion is limited to flat dithering. We have decided not to implement noise-shaping dithering because pro audio production software available on the market usually offers the user noise-shaping dithering of some kind already. We also based our decision on the fact that the sample rate conversion process often adjusts peak structure of the original program material, thus, in many cases, making a subsequent peak-limiting a necessity. To prevent output audio from clipping we have implemented a level normalization feature.
The demo version of r8brain PRO allows you to evaluate all features of the software, but its conversion is limited to the first minute of an audio file. The demo version does not support batch conversions. You may register to obtain either a Full or Light license of r8brain PRO. The Light version offers you a linear-phase conversion mode and up to 48 kHz output sample rates only while supporting all input sample rates. The Full version offers you all features of the software.
r8brain PRO features:
# Reads 8-, 16-, 24-, 32-bit PCM and IEEE files
# Reads both WAV and AIFF files
# Writes 16-, 24-, 32-bit PCM and IEEE WAV files
# Multi-channel file support
# Extensible wave format support
# EBU BWF extensions support
# Linear- and minimum-phase modes
# Ultra-steep conversion mode
# Batch support
# Automatic normalization
# Supports all standard sample rates
Last edited:
You can use a SOX plugin, too. Resampling and changing bit depth isn't the black art it used to be.
If you wanted to be super picky, you could run two simultaneous recorders, on at 16bit, the other at 24bit. You'll have a chore getting a precise level match, tho.
FWIW, I've heard a difference in the bass when upsampling to 24bit, but I suspect that was the gear in line, not the format itself.
If you wanted to be super picky, you could run two simultaneous recorders, on at 16bit, the other at 24bit. You'll have a chore getting a precise level match, tho.
FWIW, I've heard a difference in the bass when upsampling to 24bit, but I suspect that was the gear in line, not the format itself.
I am using a SOX-derived plug in in foobar2000. I have recordings that originated from an LP, recorded at 24/96. They sound absolutelly identical. Now, If I drop the resolution to 16/44.1, with my Grado headphones plugged into my E-MU 1820 I can hear the difference.
As I did say before, the source needs to have enough dynamic and bandwidth to be noticeable.
As I did say before, the source needs to have enough dynamic and bandwidth to be noticeable.
I hear the high-end sounds being "mudded", less dynamic, less "modulated". More like a constant noise instead of a fast variable one.
The High-hat, Ride cymbal, Crashes, even Trumpets sound different. All the metallic percussion in a rock band.
Also all the high-harmonics devices like electric guitar - if you pay attention even the sharp attacks sound a little different, milder, on 16/44.1. Scratched cords sound also slightly mutted...
The High-hat, Ride cymbal, Crashes, even Trumpets sound different. All the metallic percussion in a rock band.
Also all the high-harmonics devices like electric guitar - if you pay attention even the sharp attacks sound a little different, milder, on 16/44.1. Scratched cords sound also slightly mutted...
Last edited:
I don't know what reproduction devices you have and... how good is your hearing. Also, is about brain being educated to "know" how real instruments sound. Well, that is percussion, and trompets because electric guitars depend of the amp used.
But if you hunt regulary (you are from NC) and you are over 50's... slim chances
But if you hunt regulary (you are from NC) and you are over 50's... slim chances
Last edited:
one more step for the completely paranoid is to upsample the 16/44 back up to 24/96 so the computer, DAC are working with exactly the same bit rate, file size for the comparison
because it is conceivable that some playback environments will give different results do to sample rate, bit rate
of course any system with such audible sensitivity to bit rate is problematic
you could just do the equivalent pre-decimation BW reduction filtering and the word length reduction with dither to 16 bits without actually going down to 44.1 - not exactly the same as the 16/44 bottleneck but fewer steps to get right to see if either the BW or bit depth matter
because it is conceivable that some playback environments will give different results do to sample rate, bit rate
of course any system with such audible sensitivity to bit rate is problematic
you could just do the equivalent pre-decimation BW reduction filtering and the word length reduction with dither to 16 bits without actually going down to 44.1 - not exactly the same as the 16/44 bottleneck but fewer steps to get right to see if either the BW or bit depth matter
Last edited:
Well that puts my mind at ease
Mine too - I wouldn't tend to trust such marketing-heavy descriptions of resampling algorithms either
<edit> if you go here : http://src.infinitewave.ca/ then the version 5.58 looks fairly decent but 5.18 is complete junk. I wonder if the marketing description got updated too?
Last edited:
Riiight. That'll be 16 as in 256, yes?
24-16=8
2x2x2x2x2x2x2x2=256.
That was 53 posts ago.
How you expect to be taken seriously with posts like this is beyond me, a.wayne.
...and how the rest of you expect to be taken seriously when you let things like that get past you is also beyond me.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.