Additional gain with Replay Gain (dB)

[stevesound]

Hi All,

Most likely a dumb question but I'm still trying to learn, I have Replay Gain (Track Gain) set in DB Ripper , with Asset installed in my Qnap 251+ I notice in the Asset options there are 2 boxes with Additional gain with Replay Gain (dB)and Additional gain without Replay Gain (dB) Am I right in assuming if I put a value in say 5DB would this increase the volume output to my player across the board ?

Steve

[Spoon]

Re: Additional gain with Replay Gain (dB)

It would likely clip the audio.

[stevesound]

Re: Additional gain with Replay Gain (dB)

Hi Spoon,
Thanks for the reply,what are the boxes used for and if 5DB would clip the audio what value would be worth experimenting with?

Steve

[Spoon]

Re: Additional gain with Replay Gain (dB)

Realistically you cannot increase the volume there, any increase could clip.

[schmidj]

Re: Additional gain with Replay Gain (dB)

"Most likely a dumb question but I'm still trying to learn, I have Replay Gain (Track Gain) set in DB Ripper, with Asset installed in my Qnap 251+ I notice in the Asset options there are 2 boxes with Additional gain with Replay Gain (dB) and Additional gain without Replay Gain (dB) Am I right in assuming if I put a value in say 5DB would this increase the volume output to my player across the board ?"

(I apologize for the length of this reply, but the subject demands that, if you want a more complete answer,)

Well, the best way of increasing the volume of your player is to turn up its volume control. If it distorts, your playback device doesn't have enough power output capability and you need one with a significantly higher power rating.

Now regarding the gory details: Actually whether it clips, as Spoon cautions, depends. Does Asset actually change the streamed audio to add "X" DB or simply change the number from the replay gain tag? In the case of "without replay gain", it would appear that Asset actually scales the audio level of the stream. And, as Spoon says, yes, that probably would cause clipping distortion. Not a good idea at all. But it is unclear (and I'll admit, I haven't read the "help" file which might explain it) whether the "additional gain with replay gain" scales the number from the stored replay gain tag, or still scales the actual audio level in the stream. Assuming it only changes the replay gain tag value, it may still create clipping or it may not. That depends on the software and hardware of your player.

And in most of today's digital devices, it may be hard to tell absent some testing and experimenting. It has to do with how your player processes audio internally, how its own volume control works, and how it deals with the replay gain tag.

First, understand that with fixed point PCM audio, (such as exists in .WAV files and is recreated from compressed files), there is an absolute maximum level that can be encoded, called 0 dBFS where FS stands for full scale. You cannot have let us say +2 dBFS, there is no way to store such a number. Any audio level above 0 dBFS will be encoded as 0 dBFS. We call that clipping, because if you look at the waveforms of clipped signals, they look like they were clipped with a hair (or hedge) clipper. And clipped audio is distorted, possibly audibly distorted.

But we also have "floating point PCM audio" which stores an "exponent" or scaling number together with the "mantissa" or actual audio data. Given the range of the allowable exponent, it is from a practical point, impossible to clip a floating point PCM audio signal in the digital world.

However, at some point, we have to convert the digital PCM signal back to an analog voltage, in a D/A or digital to analog converter. AFAIK, they all internally are fixed point devices, supposedly subject to clipping. And here, things get a bit fuzzy. Some, not all, D to A converters do not clip, they are designed with built in limiters, which "round off" or soften the effect of the clipping to reduce the audible distortion. So technically, the audio is not clipped, although it still may have audible distortion. These built in limiters are typically not particularly sophisticated, and the audible effect may still not be pleasing. But there is more. In a typical digital home or car "receiver", almost all the circuitry is digital, including its volume control, with the D to A conversion occurring either right before the power amplifier or in some cases (such as class D amplifiers or such) even as a part of the power amplifiers. In these cases, it is possible for the clip point to actually be close to the full rated power of the amplifier or the "smoke point". This is particularly true if the internal calculations in the device (which may include the application of replay gain as well as the volume control) are done in floating point arithmetic. Here, again, there may be no clipping unless the volume control is turned way up and the output audio is very loud. But absent some fairly sophisticated tests, you don't really know what is happening internally.

And even if the application of Replay Gain occurs in a player external to the actual playback device (as when the player is in your phone, fed to your car radio or home playback device via Bluetooth), you may have the option to include a limiter in the (presumably fixed point) replay gain calculation. I'm using an Android player called Poweramp (no relation to dBpoweramp) which includes a limiter option in its replay gain application. I've never tried using it (although I may) to determine how "transparent" the limiter is, but if it is designed at all well, then adding 5 or more dB to the replay gain setting might actually make the audio sound louder without clipping or other audible distortion by compressing the dynamic range.

I don't see a similar option in Foobar2000, but someone may have written a plugin with similar functionality.

My use of the playback limiter wouldn't be to just make the audio louder, that's what the volume control on the playback device is for. Rather it is to allow the replay gain function to work better. The (default) EBU loudness replay gain calculation in dBpoweramp includes a defeatable calculation that prevents the loudest audio being above O dBFS in the player (where it is applied) in order to prevent clipping in the player. It does this by reducing the replay gain number stored in the metadata of the audio file sufficiently to keep the instantaneous audio level below clipping when the file is played. The effect is for the track to be played softer than the desired loudness that the surrounding tracks are played at. This only happens with tracks that have an (in these days) unusually wide dynamic range, but it is not that unusual in some jazz and much classical music. If you read the details of the log created by the dBpoweramp ripper, it will tell you when it did this, together with the new calculated loudness and the peak audio level of the track.

I'm considering, on the copy of my tracks loaded onto my phone that I play in the car, running the replay gain DSP with the limitation on replay gain numbers causing possible clipping turned off and the limiter in Poweramp on my phone turned on, to see how it sounds, and if it solves some of my issues with soft jazz and particularly classical track playback in my car.

[stevesound]

Re: Additional gain with Replay Gain (dB)

Hi Schmidj,

Wow! I'm sorry to say you lost me with that technical answer, for the record I use flac files for music and my DB ripper is set to Spoon's guide, I use a Linn Majik DSM 3 which is bi amp with a Linn Majik 4100 amp, and Pro Ac Studio 140 speakers, I am not sure how the Linn processes the data which is fed from Asset and Linn Kazoo but sound wise I am always very pleased, I think it sounds better than the CD played on my Roksan K3 player, with regard to the volume it is a good listening level at about 30 out of 100, I did do a test on the Asset DB boxes with 5 DB and did not notice any difference so I set back to 0, I don't hear any distortion on my DSM so I think i'm getting accurate rips, so I think you are right volume control is king!

Thanks for your interest.
Steve