Everyone should have at least some idea about what happens to their files on mix:analog.
We promised to be as transparent as possible and we're happy to honor that.
Upsampling your files
When you upload an audio file to our server, it first gets upsampled to a 32-bit floating-point, 192 kHz, .wav format.
This is required because we’re running the converters at 192 kHz to ensure maximum audio quality.
The algorithm used to upsample (SoX 14.4 VHQ Linear Phase) has been selected for minimum artifacts in 64-bit precision with linear phase filters.
A simple 128 kbps .mp3 version of the file proved to be the safest and efficient option for the quick-preview play within your 'Media Library'.
The .mp3 format is supported by all browsers and keeps the file sizes small.
The D/A conversion
The upsampled 32-bit floating-point, 192 kHz, .wav version of your file is picked up by the audio engine, played and can be volume adjusted using the 'output level' knob of the 'mix:analog IO unit'.
The audio stream is then routed to our Burl BDA8 D/A cards and converted into an analog signal.
The BDA8s have a discrete output stage for high current handling and do not employ transformers. They are as clean as D/A cards get.
Your audio then travels all the way through the signal chain to the A/D conversion stage.
The A/D conversion
The analog signal hits our Burl’s BAD4 cards, which employ big chunky nickel input transformers.
Contrary to steel or mixed input transformers, these are fairly linear and don’t saturate roughly or abruptly. Instead, they induce a softer, gradual saturation that is very much level dependent.
If driven low, they are extremely transparent and get out of the way to the discrete analog stage that sets up the A/D conversion.
The A/D conversion happens at 192 kHz and is driven by the same internal clock as the D/A conversion.
This audio stream is then routed to the audio engine and can be volume adjusted, using the second knob on the IO unit. Note that this knob does NOT soft-clip the Burl's A/D section.
When you choose a lower sample rate than 192 kHz in at the 'Bounce Settings' step, we have to downsample your audio.
As with upsampling, we are using linear phase filters and a minimum artifacts algorithm with 64-bit precision.
We also dither your audio to 16-bit or 24-bit with Shibata, low-intensity, ATH shaped triangular probability density function (TPDF).
Your files on mix:analog are handled with great care, kept secret and safe.
Our goal is to provide a secure environment and become trustworthy partners with engineers at every level of audio production.
In the future, we will also make 32-bit floating-point, .wav files available, so you may apply a choice of dither yourself.
I hope you learned something new, reading this article.
If we left any questions unanswered, feel free to mail us at firstname.lastname@example.org.
Until next time ;)