Adding dither noise to simulate SACD

boxerfan88

Well-known member
For the fun of it, started an experiment to add dither noise to simulate SACD noise.

Approach: setup a DSP chain -- after resampling to 176kHz, add SmartDitherDSP with 17 bits of content + 4bits of high passed dither.

Using DeltaWave to analyze the result, we can see the noise added in the upper frequency range (similar to SACD).
  • blue line is the original music track (Birds)
  • white line is the modified track with dither
  • I can't really hear a difference, but I can see it in the charts.

2024-03-21_2229 DeltaWave DeltaWave_v2.0.13.png
 

boxerfan88

Well-known member
When playing back music, can see the dither noise just peeking up above the floor of of the spectrum display (red box at the bottom right), so I know it's working. Theoretically, the levels are so low, it should not be audible.

Need to monitor a few days ... see of placebo effect is taking place ... and I start hallucinating that the music sounds more analog ;)

2024-03-25_0747 foobar2000 Enhanced_Spectrum_analyzer.png
 
Last edited:

boxerfan88

Well-known member
Need to monitor a few days ... see of placebo effect is taking place ... and I start hallucinating that the music sounds more analog ;)

I think I quite like the post-processing TPDF dithering to 16bits, sounds very slightly "smoother" to me.
Maybe it's placebo effect going on, I donno.
The dithering is now staying on my DSP chain ;)
 
Last edited:

boxerfan88

Well-known member
Recently changed to mda dither… even more analog…

IMG-1936.jpg




.
 
Last edited:

boxerfan88

Well-known member

Myth - "Analog tape has more resolution than digital audio." (9:37 - 11:34) Busted! Tape noise and gaussian dither noise are virtually identical. Cassette tape is equivalent to about 6 bits, and the best studio reel-to-reel tapes are equivalent to about 13 bits.

Myth - "Dither masks quantization noise." (11:35 - 13:50) Busted! Dither replaces quantization distortion with a noise spectrum of our choosing. Harmonic distortion (produced by quantization) is removed by dither and this can be seen on the analog test gear.
 

boxerfan88

Well-known member
Used my foobar to convert a sample SACD/DSD track to PCM 176k to analyze ...

Wow ... the amount of ultrasonic noise is really high.

Blue is the decoded DSD into PCM 176k.

Green is the CD version from the SACD.

2026-01-04 DSD-CD_upsampling-compare.png


Orange is my DSP "cooked" version with CD upsampling, bitdepth reduction to 18bits, and dither noise added.
The added noise is at a decently low level, not as high as SACD.

/
 
Top