Sample Rate, Frequency Achieved, Human Hearing, Distortion, and Dynamic Range Questions

[DVDdoug]

Another factor with higher frequencies is masking. Although you might be able to hear to 20kHz (or higher) in a hearing test, in context of regular music the higher frequencies are weak, your hearing is weak at the limits, and those frequencies masked (drowned-out) by slightly-lower frequencies.

One of the tricks to MP3 (or similar) compression is that masked sounds can be thrown-away. And in general when someone hears an MP3 artifact (or a difference from the uncompressed original), it's usually not the loss of high frequencies they hear.

I'm not saying you should throw-away the highest audio frequencies if you don't need to... but they are not always as important as you might think.

 

[tmtomh]

I've seen that database before. Thanks. As far as I can see what exactly do their numbers mean? So like if the database says the dynamic range max is 15 does that mean 15 bits? I didn't see anywhere on the site that exactly specified that.

Thanks

The DR meter readings in that database are based on a very particular measurement. The DR reading for a track is the difference, in dB, between the RMS (average) volume level and the second-loudest peak. I don't know why it's the second-loudest, but I presume it's to rule out any spurious individual peaks (although of course if a track has more than one spurious peak and each one is at a slightly different level, then you've still got a misleading reading).

The DR meter uses RMS as the baseline rather than the lowest level because if it used the lowest level, it would become effectively useless, and often misleading. The lowest level on any track is either going to be pure digital silence (if it's a digital recording, or if the CD mastering engineer or production house "blacked out" the between-track silences of an analogue recording), or else it's going to be the noise floor of whatever analogue system the music was recorded on.

A nicely dynamic DR12 track originally recorded on analogue tape and mastered to CD relatively quietly would read something like DR65 or 70 using such a scale, while a dynamically squashed DR5 track recorded digitally and maximized to all heck would read DR96, since its lowest level would be digital silence and it would be full of peaks at the maximum possible digital volume level.

 

[danadam]

... between the RMS (average) volume level

of the loudest 20% of the track ...

https://web.archive.org/web/2018091....de/sites/default/files/Measuring DR ENv3.pdf

 

[danadam]

material which has been recorded at a single frequency and then converted

Personally, I have no problem with that, assuming good resampler is used.

 

[MKreroo]

By looking at the DR database or using a program that determines it.

It says absolutely nothing about amps or DACs required or present bit depths or anything else.
The limiting factor is always the recording not the DAC.
When it comes to max achievable SPL (the actual dynamic range of the entire system, not the recording) the limitation is always a combination of transducer efficiency and maximum output level of the amplifier.
How loud that is to someone also depends on that persons age and hearing abilities.
Basically you just want to be sure you can reach levels you want.
I would say 120dB SPL is loud enough for impressive listening levels and the system's S/N ratio should be around 110dB.
The vast majority of recordings will be much poorer in S/N ratio so will mask system noise (depending on spectral spread of the noise and hearing abilities)

When you don't need extremely loud but only loud you can shave off 10dB or so. In all cases... the recording will be much worse than the system.

seems that I have finally found the final answer to my questions about similar topics here, really quite a goldmine here.
would this then be the answer to the long discussion on here?

Looking at the article amirm wrote some time ago, he mentioned at the end how with a peak 130dB recording and 0 dB noise floor, it translate to 130 dB DR. I assume this isn't just a simple 130-0, but that since the noise floor is essentially below threshold across all frequency, there's no way for listener to hear anything below the noise floor.

I personally have some pieces that have very low DR (5) that I found good. Don't have problem with this as here it has been discussed that compressed DR can sometimes be good, and that I enjoy it, just confused if it is possible for a piece that has intended background audio (not necessarily noise) throughout the entire track, thus having low DR? or that I just don't mind low DR.

thanks

 

[solderdude]

A recording can have a low DR and still sound good. This totally depends on the music itself and how good the mixing engineer is in using the tools he has available.