This issue is killing me...

  • Thread starter Thread starter ambi
  • Start date Start date
A

ambi

Guest
Hi all,

First off, I'm new here and I've decided to join mainly because of my problem that has been bugging me for a while. I have been a passive forum browser beforehand and the forums have been extremely helpful!

Anyway, back to the main issue...

I've been having audio problems when uploading my wav file to Soundcloud, or converting my wav to an mp3. Pretty much when playing my sample through Ableton Live its clean and crisp sounding, and is just as high quality when exported as a wav file, however, when I convert from wav to an mp3, either using Audacity or Switch, the audio quality has degraded and there are a lot of sound artifacts. I've been testing this over and over with a kick drum loop I've made myself, and every kick hit seems to have a clicky/glitchy/artifact sound as an mp3 file and on Soundcloud.

I am aware that mp3s and Soundcloud playback is lower quality than a wav file, but big producers out there manage to have a clean sounding mix, even when playing back at 128kbps on Soundcloud.

I have checked my Ableton preferences
Default SR and Pitch Conversion: High Quality
Sample Rate: 44.1kHz
Buffer Size: 512 (Tested other buffers, this was best)

My kick drum sample is 16bit 44.1kHz, I've put it in a Sampler Instrument, and made sure the Interpolation settings are at Best. I've EQed the sample and made sure the EQ is High Quality, Hi-Passing around 40Hz and Low-Passing around 16kHz. I've used a compressor. There is no clipping whatsoever. Finally, I've got an Izotope Maximiser at the end of the chain, setting the margin to -0.5db, which I assume is plenty of headroom.

When I render
Sampling Rate: 44.1kHz
Bit Depth: 16bit
Dither: Triangular

I've tried all sort of combinations, removing devices, changing preference settings, changing render options such as dither on and off, even trying different kick samples, and then exporting and checking as a wav and mp3, and still I get bad quality mp3s...no success.

If anyone could help it would be much appreciated.
 
whatever you are using to convert to mp3 make sure that it does not have a normalise step in its processing chain - this can make small problems bigger and big problems bad.

Secondly only ever upload a wav to soundcloud - no need to have them re-encode from your mp3 to their mp3 (they will do it regardless of file format (re-encode to mp3 at their end), so upload the best quality you can) remember they measure your account in minutes not in mb or gb
 
Last edited:
I'll have a check but I don't think Audacity is normalising my files.

I always upload a wav to Soundcloud.
 
Maybe you could post one of your problematic wav file here?
 
ambi said:
I'll have a check but I don't think Audacity is normalising my files.

I always upload a wav to Soundcloud.
Click to expand...

in the standard install - under file menu->edit chains->MP3 conversion

the standard chain is

normalise to 0db
export mp3
end
Click to expand...

You need to remove that first step in the chain, otherwise your audio gets mangled and loses dynamic range before it is encoded to mp3. as the mp3 codec relies on removing certain areas that supposedly are masked or shadowed by louder events, you are creating the potential for greater conversion artefacts by leaving the normalise in the chain.....
 
There are a few things here that I feel might be causing your problem. First, your session sample rate is at 44.1. You need to change that to at least 48. The higher the quality of your session, then cleaner it'll sound when you bounce it down. Also your kick drum sample should be higher quality. You should only be using 24 bit WAV files for your drums. Third, when you export your track, also export it as a 24 bit WAV file. the Only time you should be in 16 bit is after you convert your 2 track 24 bit wave file into a 16 bit, 44.1khz mp3 file. You might want to try a new mp3 converter too. maybe All2mp3(the one i use).I've never seen a Triangle Dither Option but that sounds like it could also cause some popping and clicking issues. Try different options and see how that changes the sound. I hope this helps
 
makes no difference what the session sample rate is - when you convert to mp3 all bets are off.

mp3 is never 44.1kHz 16 bit; at best you get CBR320kbps.

I'm trying very hard not to diss you but your overt lack of knowledge is making that difficult.

I have heard better sounds using an 8 bit mono sampler and a 12 bit mono sampler than some of the so-called pro sounds at higher bit depths and higher sample rates - there is nothing inherently wrong with using a sample rate and bit depth that is appropriate to the sound you are trying to create.

Session sample rates are tied to the maximum frequency range that you can record.

Sample rates have no impact on quality once you hit the 44.1kHz mark - human hearing range is 20Hz to 20kHz more or less.

The sample rate needs to be higher than twice the highest frequency to be recorded to avoid artefacts known as Nyquist foldover or Nyquist noise, where the frequencies just about the half-way point (the Nyquist point) of the sample rate are reflected back across the Nyquist point and are heard as high frequency hash or noise.

Most audio cards deal with foldover noise by applying a low pass filter with its corner set 20kHz. All higher sample rates are about oversampling to provide more slices or greater clarity.

Bit depth on the other hand does matter as it is a measure of the dynamic range able to be stored.
8 bits is approximately 48db dynamic range,
12 bits is approximately 72db dynamic range,
16 bits is approximately 96db dynamic range,
24bits is approximately 144 db dynamic range,
32 bits fixed point is approximately 192db dynamic range.
 
Last edited:
The sample rate of the session does matter, or else why would it even be there. If there were no reason to have a 96 khz sample rate session or a 192 kHz sample rate session then it wouldn't exist and everything would be recorded at 44.1

The higher the sample rate the more times your audio is being sampled per second, there for making you audio sound cleaner as you stated. Then when you bounce it down, more of the information is retained. Its subtle but you can hear the difference in your hi's. Go ahead and test it out for yourself with a PAZ Analyzer. And mp3 does come in 16 bit 44.1 kHz. It's the equivalent of 176 Kbps but it gets brought down to 128 Kbps.

You don't to talk to me about the nyquist theorem and the range of human hearing because I already learned all that stuff.

And your point about 8 bit or 12 bit sounds verses so called "professional" sounds is irrelevant because there are so many more factors that directly effect how good a sound might be. If you you take a 24 bit sound, compress the living shit out of it and throw some horrible eq on it than you may just find an 8 bit or 12 bit sound that sounds better than that.
 
Tru-theProducer said:
The sample rate of the session does matter, or else why would it even be there. If there were no reason to have a 96 khz sample rate session or a 192 kHz sample rate session then it wouldn't exist and everything would be recorded at 44.1

The higher the sample rate the more times your audio is being sampled per second, there for making you audio sound cleaner as you stated.
Click to expand...
I did not state that it makes your sound cleaner or clearer.

Sample rate is purely abut increasing bandwidth - it has absolutely nothing to do with a sound being cleaner or clearer - any harmonics above the range of human hearing are not worth recording as they cannot be perceived on playback, nor do most systems have the ability to reproduce such a wide frequency range (96kHz @ 192kHz sampling rate).


Then when you bounce it down, more of the information is retained. Its subtle but you can hear the difference in your hi's. Go ahead and test it out for yourself with a PAZ Analyzer.
Click to expand...

I think you are believing some hype rather than stating your own experiences. Also please note that I am not saying that if you can't hear it it isn't doing anything, far from it.

There can be no quantitative improvement in the conversion of a high sample rate to the lower output sample rate of the soundcard involved in the process - the limits are in the dac's and there will be a loss of information rather than a retention of it, mainly because of the LPF with it's corner set at 20kHz.

The qualitative aspect is purely subjective - everyone will experience something different

And mp3 does come in 16 bit 44.1 kHz. It's the equivalent of 176 Kbps but it gets brought down to 128 Kbps.
Click to expand...

it's not, simply because in the process of encoding an mp3 there will be data loss - mp3 codec is a lossy compression technique that removes masked or overshadowed frequencies based on a moving Hamming windowed FFT.
 
I <3 nyquist fold over

---------- Post added at 08:05 PM ---------- Previous post was at 08:05 PM ----------

Makes your shit sound like whales in love.
 
Do you think you guys could analyse the files?

Here's a hi-hat loop I made as a wav file

Zippyshare.com

This is the same loop converted into a 128kbps mp3 file

Zippyshare.com

You can here obvious glitch sounds.. I think it may be the wavs fault even though I can't here the glitches in the wav file
 
You must log in or register to reply here.
Back
Top