Is 32 bit audio better?

August 19, 2021
 by Paul McGowan

9 comments on “Is 32 bit audio better?”

  1. The explanation of the dB scale was fair but there was no explanation of what sample rate is!

    Iā€™m not sure many can explain either measurement even though they are used constantly!



  2. Paul,
    Could you give me a description of the room you used for this video. It is obviously acoustically prepared due to the speakers used, but also, your voice sounds very natural in it. What treatment was used? Just curious. (This could be a whole video in itself.)

    1. That’s Music Room II and where we do all of our reference listening to both equipment and Octave releases. It has minimal acoustic treatment. We designed the room from scratch with the proper proportions, angled the walls, and then placed Vicoustic abfusers on the sidewalls at the point of first reflection (you can see them behind me).

      There’s some treatment in the rear of the room, but mostly it’s left untreated because it sounds quite natural and slightly live.

  3. I think the question has two answers: (a) what’s an appropriate bit depth for recording and publishing audio data and (b) what bit depth should we use inside a digital audio workstation.

    Here’s a worked example for the output from a microphone, assuming a sensitive microphone (5 mV/Pa), 150 ohms source, 20 kHz bandwidth, a loud 110 dBA sound level, and a pre-amp with a 1 dB noise figure. The signal to noise ratio is about 102 dB. Now consider that we want 10 dB headroom at the top end, and we’d ideally like the ADC quantizing noise to be 10 dB below the thermal noise. Now we need 122 dB range. That’s equivalent to 20 bits. Clearly there’s a theoretical benefit in using more than 16 bits. On the other hand, 24-bits is not needed and, in practice, not achievable. Still, engineers like multiples of 8 bits, and so we have a standard for 24-bit audio. There’s absolutely no point in going beyond this.

    It’s a different consideration inside the DAW. We’d like to remove any worries about dynamic range inside the DAW, so that we don’t have to avoid overload or accumulating noise from rounding errors. Here it makes perfect sense to use 32-bit processing. It costs nothing in processing power and it makes the workflow so much simpler.

  4. The Patricia Barber album Higher was recorded, mixed and mastered at 352.8 / 32 bit. The SACD was authored by Gus Skinas. The stereo version is very good, the multichannel is a sonic tour de force. This is the best sounding SACD in my library by a wide margin! The recording engineer, Jim Anderson says in the liner notes that he feels he has created a new audiophile standard by using DXD for recording, mix and master. I would invite Paul to listen to Higher and then see if he still thinks 24 bit is good enough.

    1. You mentioned that your copy of Higher is on SACD. I guess this is this encoded as standard DSD64. If so, the dynamic range on replay is 120 dB or about 20 bits. As I tried to show in my comment above, 20 bits is about the practical limit for any recording system.

      I don’t doubt that it’s a wonderful album and a new audiophile standard, but there is no way to record music with a 32-bit range. It does, of course, make sense to process the recordings with 32-bit precision in the Pyramix system before creating the DSD version for consumers.

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram