Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Dithering Audio / Hearing Damage (Read 5965 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Dithering Audio / Hearing Damage

Hi, in an attempt to learn more about the mathematical details of dithering I happened to stumble across Naoki Shibata's SSRC project which has a great dithering algorithm in it.  I was very impressed with the low amount of noise with his algorithm when using dithering when going from 16 bits to 8 bits.

However, I found something in his notes which peaked my interest.  He writes:

Quote
Dithered 8bit files contains strong supersonic, and listening to these
files for long hours may damage your hearing. Dithered 16bit files are no
problem since the power of the supersonic of dither is 1/65536 of those of
dithered 8bit files.


...I had never heard of this before.  I searched the web some looking for info concerning dithering and hearing damage but found nothing.  So I emailed Naoki asking about it but the email address he provided is not valid.  I've searched the web to find another way to contact him but had no luck.  Does anyone else know anything about hearing damage caused by extended listening to audio that has been dithered or know where I can read more about this?  Thanks.

Dithering Audio / Hearing Damage

Reply #1 –
This is something I had discovered when I was exploring dither.

Dithering a 32 or 24 bit file to 16-bit, when done with a good noise shaping, effectively makes the dither inaudible in the vast majority of cases.  So in order to train myself to better understand dither, and more importantly to hear it, I would dither my 24-bit recordings down to 8-bit.  With no noise shaping, you just have strong white noise everywhere.  With a strong noise shaping, you will have a very high-level noise content in the 18-20khz area (assuming 44.1khz samplerate).  If you can hear these frequencies, they will feel piercing and cause pain at high enough volumes.  That should be enough for you to avoid them.  The real danger comes in when you cannot hear those frequencies.  They can still produce hearing damage, in addition to massive headaches and other various unpleasant things.

What I usually do, if I actually want to use 8-bit files for music, is dither them down, and then apply a 16 or 17khz lowpass filter.  Problem solved!

Dithering Audio / Hearing Damage

Reply #2 –
OK, now I'm confused. Wouldn't the digital filtering simply reintroduce the quantization noise that you had just dithered out?

Dithering Audio / Hearing Damage

Reply #3 –
Quote from: pdq on
OK, now I'm confused. Wouldn't the digital filtering simply reintroduce the quantization noise that you had just dithered out?
Yes, it will. The noise will be shifted back into the audible band by the filter.

Is this really worth worrying about? What sort of application do you have in mind for 8bit linear PCM anyways? If it's just a casual interest, then read up on how dither works (several good articles linked from the FAQ), then realize that because of the threshold of hearing you won't hear this huge amount of high-frequency noise. If the content played loud, this noise might be loud enough to cause hearing damage.

Dithering Audio / Hearing Damage

Reply #4 –
Quote from: cabbagerat on
What sort of application do you have in mind for 8bit linear PCM anyways?
Kid's toys come to mind. Our kids have plenty of those horrible sounding toys. Luckily it's very unlikely that the high frequencies of noise shaped dither well ever make it through the tiny 10 cent speakers . From what I've read toys audio sample rate is typically in the 6kHz to 8kHz range. Apparently 4 bit audio was not unusual. I have the impression that toy sound quality is gradually improving. (info)
Several studies indicate that sound levels produced by baby/kid toys at 1 cm (babies often put the toy to their ear) can exceed 100 dB spl. I can imagine that at those levels, hearing damage can not be excluded, with or without dither.

Dithering Audio / Hearing Damage

Reply #5 –
Thank you everyone for the feedback.  I'll check out the FAQs on this right after this post.

Quote
Is this really worth worrying about? What sort of application do you have in mind for 8bit linear PCM anyways?


Really this is a personal concern.  Using 8 bit audio makes it A LOT easier to hear noise introduced when downsampling from 16 bit as opposed to downsampling from 24 bit to 16 bit.  So, when experimenting with dither it is very nice to work with 8 bit. 

The last thing in the world I need is to loose any more hearing ability.  I'm 32 and given the volume that I used to listen to music and play in guitar in bands I have noticed hearing loss.  Being in a bar or restaruant with background noise it is very hard to have a conversation - the people I'm talking with get drowned out when other people in the conversation don't seem to have a problem.  Just trying to avoid getting a hearing aid before I'm 40 

So, this may sound like a total newbie question (so I apologize before hand) but with these techniques can my hearing be damaged if it doesn't physically hurt my ears at the time the audio is being heard?

Terence