Fascination Street says reamp/di signal is too low.....

Status
Not open for further replies.
You are confusing bit-rate with bit-depht .

Actually bit-rate is called sample rate Fs and is messured in Hz (cycles per second)

Bit depth (resolution) is messured in Bits

Clarky that is exactly what I mean whatever bit rate a digital sample is it is invariable. If it is 16 bit its 16 bit, if its 24 bit its 24 bit.

However I am asking Thomas, does he think by volume lowering bit rate changes as that is what he seems to be saying.
 
Actually:
Bit rate - amount of data per second, refers to lossy compressed audio formats, like MP3
Sample rate - how many times sound is sampled per second
Two different things, mp3 can be 44.100 Hz/s and 320bps
Bit depth - dynamic range of the sample in digital realm

To simplify things In non ideal world dynamic range is determined by headroom (ceiling) and noise level (floor). I'm pretty sure that 24 bit is way beyond any electric guitar's dynamic range.
 
The point was :

Let me take an extreme example.

If you was to capture a signal (in 24-bits) that is so weak that it only touches the first digital-bit, you will actually record one bit (changing from 0 to 1) and 23 zeros (won´t move). If you then normalize the sound-file ,you will still have a signal that has the quality of a 1-bit sampling even if it is represented by 24-bits (actually a 100% leveled 24 bit square wave).

If this extremly low signal was an triangle wave from the beginning it would then be destoyed even if it was normalized (amplified)
Try and take a photo of a chess board with one pixels resolution and than magnify it and you will see what I mean .

I know these examples are extreme but hopefully it explanes what I mean.

You are confusing bit-rate with bit-depht .

Actually bit-rate is called sample rate Fs and is messured in Hz (cycles per second)

Bit depth (resolution) is messured in Bits
 
Thomas, now I see where you are coming from. Yes the more times a second the fs are and the more bits used to sample each occurrence, it would of course result in a more detailed digital representation, than a lower bit depth and sample rate.

E.g. telephones, 8000hz * 8bits = 64Kbits, but those rates are invariable. The sample still occurs 8000 times per second and sampled at 8 bits each time. This happens regardless of whether there is silence or not. When this is transported the bit stream of 64Kbits has to be maintained so if there is silence, bit stuffing occurs.

But perhaps you can explain more why you think if the signal going in is hotter there is more clarity if the bit depth and the bit rate does not vary ?

Sampling rate - Wikipedia, the free encyclopedia
 
The point was :
If you was to capture a signal (in 24-bits) that is so weak that it only touches the first digital-bit, you will actually record one bit (changing from 0 to 1) and 23 zeros (won´t move).

What Cliff is pointing out is that self noise of guitar is much louder than "microinformation" that you could capture using only full 144 dB of dynamic range, so it would be completely covered by noise, same as when you don't hear people whispering during loud concerts. The floor for dynamic range in real world is always noise, not perfect silence.
 
I'm not sure if you're replying to me or not but...
no.. you'll always be using 24 bits...
You're misunderstanding me. "Uses all bits" means using the range 0x0-0xFFFFFF. :)
Thomas is concerned that a low level will squeeze the detail out of the recorded signal..
if the level is extremely low then yes this can happen...
but I think that this is not the case for what is being discussed here because the signal level falls well within the limits that could cause issues..
Right. I think the real issue is s/n ratio and nothing more. The whole talk about bits using "must have been done for a reason" as an argument and then assuming the reason has to do with input level, doesn't make sense to me.

When I mentioned upsampling, I was referring to what is done internally in many high end units when doing heavy audio processing to avoid losing precision.
Downsampling is then done at a later stage before the output. The OP is however talking about DI levels, hence I figured that was out of the scope of the thread.
 
We went from 16 to 24 of a reason. I'm gonna use them all .
Period !
(Do what ever you want)
Check how sound engineers record.
With high gain sounds we amplify the guitarsignal
100000 of times. We better get the lowest levels right before we magnify them that much.

:)

I am a sound engineer. I know how we record, and I know why. You're just guessing ;) What those high gain sounds magnify is noise. It has nothing to do with bit depth, as those bits represents a wider range than the signal itself. Even at -40 dBFS!

The reason we now record at 24 bits is to NOT having to push everything, and use compressors/limiters, on the way in, and we have more room for number crunching/processing. On the Axe you push the input to get the best possible s/n ratio for that tiny signal which is to be amplified so much, but when you've entered the digital realm the lower bits are just pure noise.

By the way , how can you possible know how much "dynamic range" my ears need ?
It's just a bunch of numbers.

It's not about your ears, it's what the electric guitar signal represents. If it is e.g. 90 dB from peak to pure noise, that's what it is. Makes no difference if those 90 dB's go from -91 dBFS to -1 dBFS or -110 dBFS to -20 dBFS.

Of course there are high slope lp-filters in the converters that takes away the
"chess board effect" on the signal. But still , why not use all "pixels" ?

There are no "pixels" or less detail! If the bits represent more than the dynamic range of a given sound the extra bits will just capture noise, or nothing. It's not like those extra bits are crammed into the signal portion of the recorded material, making it less "pixelated".
 
If you was to capture a signal (in 24-bits) that is so weak that it only touches the first digital-bit, you will actually record one bit (changing from 0 to 1) and 23 zeros (won´t move). If you then normalize the sound-file ,you will still have a signal that has the quality of a 1-bit sampling even if it is represented by 24-bits (actually a 100% leveled 24 bit square wave).
In the Axe-Fx you are ALREADY hitting the converters full scale, then the reamp feature gives you the exact digital representation of what's the output of the input AD converter.

Routing the DI signal then boosting it and record the boosted version will NEVER capture more information, because you are falling in your own exaple quoted above.
 
Well It seems we are hitting the coverter correct
but it does not get stored in my DAW at full scale of som reason.
The must mean that the last bits is thrown away.

In the Axe-Fx you are ALREADY hitting the converters full scale, then the reamp feature gives you the exact digital representation of what's the output of the input AD converter.

Routing the DI signal then boosting it and record the boosted version will NEVER capture more information, because you are falling in your own exaple quoted above.
 
Well It seems we are hitting the coverter correct
but it does not get stored in my DAW at full scale of som reason.
The must mean that the last bits is thrown away.

IMHO you are drawing too many conclusions based on things you don't fully understand.
The level as you see it in your DAW has nothing to do with number of bits, A/D conversion or input level. You won't get more volume with 32 bits or less with 16.
If it makes you feel any better, think about this: When you use the most significant bit of the 24 bits, you are using _all_ the bits, but still have half the way to go before you reach the maximum value.

As for getting more volume into the DAW, have you thought about increasing the output level in your presets?
 
I think you guys are not getting what Thomas is talking about.

Maybe a drawing comparing a triangle-wave captured at full level and captured with a really low level would help.

I think the issue here is quantization noise.
 
Last edited:
I
It's not about your ears, it's what the electric guitar signal represents. If it is e.g. 90 dB from peak to pure noise, that's what it is. Makes no difference if those 90 dB's go from -91 dBFS to -1 dBFS or -110 dBFS to -20 dBFS.

Exactly. A typical guitar has less than 100 dB of dynamic range. A typical guitar peaks at around -20 dB re. FS on the Axe-Fx II. The Axe-Fx II outputs 24 bits which is 144 dB of dynamic range. Therefore, if the guitar is peaking at -20 dB re. FS, there is 124 dB of dynamic range left. Since the guitar dynamic range is only 100 dB, then the bottom 24 dB is noise.

Or you can look at it this way: a typical guitar has only 16 bits of usable information. Since it peaks at -20 dB then the upper 3-4 bits of the Axe-Fx aren't being used (let's say 4 bits are being "wasted"). So there are still 20 bits left. Since the guitar only has 16 usable bits then the bottom four bits are noise.

Now, depending upon how your DAW translates from 24 bits to it's native bit depth may make it seem that the signal is very low. If it maps those 24 bits onto the bottom 24 bits of a 32 bit word then it will appear that the signal peaks at -72 dB (12 * 6).

The technique that the Axe-Fx uses is accurate, reproducible and saves ALL the information from the instrument. To suggest otherwise represents a lack of understanding in how the process works.
 
Exactly. A typical guitar has less than 100 dB of dynamic range. A typical guitar peaks at around -20 dB re. FS on the Axe-Fx II. The Axe-Fx II outputs 24 bits which is 144 dB of dynamic range. Therefore, if the guitar is peaking at -20 dB re. FS, there is 124 dB of dynamic range left. Since the guitar dynamic range is only 100 dB, then the bottom 24 dB is noise.

Or you can look at it this way: a typical guitar has only 16 bits of usable information. Since it peaks at -20 dB then the upper 3-4 bits of the Axe-Fx aren't being used (let's say 4 bits are being "wasted"). So there are still 20 bits left. Since the guitar only has 16 usable bits then the bottom four bits are noise.

Now, depending upon how your DAW translates from 24 bits to it's native bit depth may make it seem that the signal is very low. If it maps those 24 bits onto the bottom 24 bits of a 32 bit word then it will appear that the signal peaks at -72 dB (12 * 6).

The technique that the Axe-Fx uses is accurate, reproducible and saves ALL the information from the instrument. To suggest otherwise represents a lack of understanding in how the process works.

What about the noise floor of the Axe-Fx? If the noise inherent to the Axe-Fx is higher than -144dB, then you "lose" some "usable" range to noise. The spec sheet even admits ">110dB" of dynamic range. It's a bit misleading to say that you have a full 24 usable bits. In the digital realm, sure. In the analog realm, it's no simple matter to get the noise floor that low.

Regardless, the noise floor is low enough that it's dead quiet through very sensitive IEMs and compares favorably to my dedicated headphone dac/amp set up (including any noise introduced in the FX's headphone amp circuit). Point is, a guitar would never be able to use even close to 24 bits of data even if the Axe-Fx were also able to.
 
What about the noise floor of the Axe-Fx? If the noise inherent to the Axe-Fx is higher than -144dB, then you "lose" some "usable" range to noise. The spec sheet even admits ">110dB" of dynamic range. It's a bit misleading to say that you have a full 24 usable bits. In the digital realm, sure. In the analog realm, it's no simple matter to get the noise floor that low.

Regardless, the noise floor is low enough that it's dead quiet through very sensitive IEMs and compares favorably to my dedicated headphone dac/amp set up (including any noise introduced in the FX's headphone amp circuit). Point is, a guitar would never be able to use even close to 24 bits of data even if the Axe-Fx were also able to.

The dynamic range of the Axe-Fx is greater than that of a guitar. Since you can optimize this dynamic range using the Input Trims, then it is not a misleading statement.
 
The dynamic range of the Axe-Fx is greater than that of a guitar. Since you can optimize this dynamic range using the Input Trims, then it is not a misleading statement.

On that point, I don't disagree. Relative to guitars, the Axe-Fx has more than enough overhead. I'm just talking from a more "absolute" standpoint. It doesn't really matter anyway since the Axe is still pretty damn good in that context as well.
 
On that point, I don't disagree. Relative to guitars, the Axe-Fx has more than enough overhead. I'm just talking from a more "absolute" standpoint. It doesn't really matter anyway since the Axe is still pretty damn good in that context as well.

That is the whole point. The digital bit depth on the USB and Digital I/O exceeds both the dynamic range of the Axe-Fx itself and certainly that of any guitar. Furthermore the bit depth is sufficient to fully capture the dynamic range of a guitar while still maintaining +20 dBu as full-scale.

At this point I am considering this discussion closed and locking the thread because people are spouting opinions which are totally contrary to science, math and facts.
 
Status
Not open for further replies.
Back
Top Bottom