I’m looking for an answer a mere mortal can understand. I am not a sound engineer. I am, however, a music geek. I’ve started digitizing some LP’s again. Here’s the pipeline:
Project Debut Carbon turntable ==>> NAD PP4 USB phone preamp ==>> M2 Mac Mini ==>> Audacity. I record into 24 bit AIFF. I then export the AIFF and import it into VinylStudio where I split it into tracks and clean up clicks, hum, and rumble. I then export it into Apple Music in Apple Lossless format. Origin syncs from the Apple Music folder. I play Origin into my tube amp via a Schiit Modius DAC. I love the quality of the sound.
I recently digitized a mono LP of Mozart piano music. The LP is circa 1957. Yes, 1957.
The first time I played the album in Origin I forgot to run Recompute Gain. The sound was kind of muddy/staticy. After I ran Recompute Gain the sound cleared up dramatically. Here’s my question: how come running Recompute Gain improved the sound so dramatically?
The ‘Gain’ of a track is relative to the loudest peak in the recording… files can be digitally adjusted to maximize the dynamic range of the recording through DSP (Digital Signal Processing) relative to the loudest peak in the recording (called “Normalization”) which is what Replay Gain is applying to the file(s) depending on the chosen setting in the Replay Gain function.
The only reason I can think of, is that enabling ReplayGain makes Audirvana apply its own volume change to the track (like the software volume control function). Audirvana does it by upsampling the track to prevent aliasing (if I remember well, it’s an upsampling followed by a downsampling operation, but don’t quote me on that). You might see the output showing a different frequency whether you enable ReplayGain / software volume or only allow hardware control. Depending on your setup and likeness, upsampling can produce various results, and that involves ReplayGain too, it’s not neutral.
Audio normalization is the process of measuring an aspect of an audio file and adjusting it so that it matches a predefined target. Learn why audio normalization is used and how to normalize your audio for different playback scenarios. https://www.izotope.com/en/learn/audio-normalization.html
First of all, thanks for your responses. What I’m taking away is that Recompute Gain should affect perceived volume but should not affect audio quality. There must have been something else going on. Per usual, I will never know. Thanks again for your responses.
I suspect that you are not using ‘limiting’ in your A/D chain, so it must be very difficult to get a good SNR (Signal to Noise Ratio) from the LP of any recordings with large volume swings…
Using a different ADC (Analog to Digital Converter) Computer interface device with high sample-rate capability, will allow you to digitize your LP collection at sample-rates up-to 176.4kHz or 192kHz or even 1-bit DSD… A limiter will allow you to capture the vinyl signal with the best possible SNR and sample-rate resolutions without getting digital clipping (overs) distortion.
A simple recording axiom: “Garbage In… Garbage Out”
I just looked up CHANNEL D - Pure Vinyl and the last OS they support is Catalina. I’m loathe to learn yet ANOTHER piece of recording software. To my ears, a 50 year old LP (sometimes mono) sounds pretty good after Vinyl Studio is finished its audio clean up. I guess each of us has our own definition of how good is good enough.
Utilizing a limiter will allow your NAD ADC to maximize the SNR (Dynamic Range) that you capture from your LP’s… There will be a fine balance to be found when using limiting, so to retain the musical dynamics of the source recording… Technically, not a trivial endeavor… It is a tedious process of finding the peak signal(s) per track and setting the input gain so to maximize without squashing the dynamics… https://www.izotope.com/en/learn/compressor-vs-limiter.html
You may want to further explore the normalization and the Software Recording Equalization features of Vinyl Studio… After capturing the source dynamic range in the digitization process utilizing limiting, Vinyl Studio should always produce a high quality file for playback… https://www.izotope.com/en/learn/digital-audio-basics-sample-rate-and-bit-depth.html
Well, I said the exact opposite (although I’m not talking about “quality”, it’s not really the topic), maybe my answer was clumsy, I’m sorry.
If you let Audirvana control the volume of your output (whether by enabling software volume control OR ReplayGain), Audirvana will definitely do something to your signal before sending the PCM stream to your DAC.
Most devices have a volume control, either analog or – increasingly – digital. When this setting is controllable by Audirvāna, it is visible with the volume bar in the software, and you can adjust it thru the application.
If this is not the case, you can always adjust the volume outside of Audirvāna using directly the interface or volume knob of your device.
However, Audirvāna offers a very high quality internal digital volume control. It is disabled by default, but you can enable it in the software’s audio settings.
In this case, Audirvāna performs the attenuation calculations before sending the data stream to your device. However, it is preferable to only use one at a time. If you use the Audirvāna internal volume control, it is best to set the volume of your device to its nominal volume, i.e. the volume at which it neither attenuates nor increases (= 0 dB).
Please note that this feature is only available when your device is connected via USB. When connected via UPnP/DLNA the large buffer creates a latency incompatible with this service (time lag).
There’s an explanatory video on the page.
There’s also this short post from the staff, if you still think RG doesn’t affect audio : Replay Gain bit perfect? - #4 by Antoine
@Yohmi , @TribecaMikey
Anytime the bit-signal is decimated to reduce level (amplitude), or the level of the file is increased by a digital volume control DSP, it is changed… The results are relative to the number of bits in the result of the digital volume control DSP (6db = 1-bit) … In the case of attenuation these bits are gone… however the loss of bits does not affect the sample-rate resolution of the file, just the dynamic range… generally, dithering will supplant the missing bits… In the case of increasing the volume, bits are added… Replay Gain is ‘normalization’.
@Antoine_G . Back in 2020 you wrote “using ReplayGain modify the audio signal and you’ll need a step of 6db if you don’t want to loose the bitperfect capability.” Two questions:
In the case of the file you are playing in the screenshot… We see this is a 24 bit/44.1kHz recording and a positive replay gain factor of 0.5dB is being applied before output… We see that the up-sampling DSP processing is producing a 32bit/192kHz file for output to your DAC… The -6dB attenuation is giving your DAC some headroom before clipping of 1-Bit (31)… a 32-bit PCM signal has a theoretical Dynamic Range of 192dB [ (6dB/1 (bit) x 32 (bits) = 192dB ] … This is unrealistic, as DACs today will not reproduce the theoretical dynamic range of the file, However, the resulting dynamic range of the file will have impact on the behavior of the DAC circuit topology and clipping distortion can be precipitated if the gain of the signal is not managed properly…
A 24-bit PCM file will have a theoretical dynamic range of 144dB and a 16-bit PCM file will have a theoretical dynamic range of 96dB… The general maximum SNR of a vinyl LP is approximately 68dB, however, this is different than the playback dynamic range of the signal reproduced by any given MM or MC cartridge/stylus, which will easily be captured by the dynamic range of a 24-bit ADC… The qualitative differences will generally be found in the resolution of the ADC (44.1kHz, 48kHz, 88.2kHz, 96kHz, 176.4kHz, 192kHz… etc)
*Note:
The upsampling to 192kHz does not add new harmonic information to what was captured in the original ADC sample-rate… The upper harmonics are intrinsically tied to the Nyquist Frequency of the ADC… in the case of the 24/44.1kHz file, the upper limits of the ADC is limited to approximately 22kHz with a brick-wall Nyquist (Fs) filter… so even if the source (LP) recording contained more harmonic information found above 22kHz, it has been eliminated along with contextual relationships to the symbiotic subharmonics…
This is why higher ADC sample-rates are capable of capturing more contextual information from the source, as the Nyquist Fs is pushed farther out as sample-rate increases and more samples are acquired.
**Note:
Upsampling to 192kHz produces a signal to your DAC output circuitry that is more refined and closer to that of an analog waveform… as sample-rate increases the waveform becomes more refined… and in the case of DSD sample-rates, the waveform is extremely close to the shape of the analog signal waveform.
***Note:
In the context of converting your LP collection… A high sample-rate ADC will capture the relevant contextual information encoded in the vinyl… I presume NAD has determined that 44.1 or 48kHz represents a qualitative sample-rate that is satisfactory for a large demographic of users of this device, as they have presumed that a Nyquist Fs of approximately 22 or 24kHz is sufficient for most… However, this does not address the resolution of the capture, which is sample-rate dependent… with the increase in samples, the better the resolution of the capture.
@Agoldnear, thanks so much for taking the time to write your response. I am working my way through it slowly. Just for the sake of completeness, here are the specs of the NAD PP4 ADC. I have an MM cartridge.
