the NSA selector

what is it

the NSA selector is a eurorack module with two ethernet jacks and one audio output. any bit on the network will be sent to the audio output.

what it's not

this is not an "audio interface". we do not play back any "format" such as RTP or MP3 or WAV ot the like. the eurorack module does not "speak" any protocol. all traffic is forwarded from one network jack to the other unmodified. it's just tapped, intercepted to convert it to audio.

examples

watch the NSA selector video

sequencer script

in the folder sequencer/ you find a very simple shell script, that mimics a sequencer by network pings of different size.

plain image transfer

if we transfer uncompressed, unencrypted images e.g. in the .bmp format, we can hear the pixels. together with a small http server which is available in the fileserver/ folder. you can listen to your photos or drawing from gimp (or photoshop in case you're a rich musician).

encode audio to NSA's native format

the NSA selector's native format is 4 bits and 25MS/s which originates from the typical PHY MAC interface called MII.

at first glance 4 bit audio appears to be really crappy, but we can use the ridiculously high sample rate. what we need is called a delta-sigma modulator. this lets us convert a simple mono 16 bit 48kHz .wav file to a 4 bit 25MHz .nsa file.

note that this saturates the link and through the added headers from ethernet, IP, UDP or TCP and HTTP you'll get artifacts. and happy litte retransmissions.

far from HiFi quality, but the method adds a lot of spice and excitement.

there's a converter in the upconverter/ folder.

network overhead

here's what a network packet can look like on the wire:

and we're listening in on the "4B" side of the "4B5B encoding" layer. so the first bits we hear are the preamble of the ethernet frame and we follow up the stack. e.g. ethernet, IP, TCP, HTTP, BMP.

delay, oversaturation

there's a neat little idea I had during development to add a delay to say a sequencer pattern. actually it's more like an echo than a delay since it lacks precise timing control. ssh log into the remote machine that creates the e.g. ping traffic. then dump the network traffic to the text console to double it. you can increase verbosity levels of the dump to crank up the echo until total saturation of the link and lost captured packets. see the video if this explanation is not clear to you.

commands I usually use:

tcpdump -ni eth0

tcpdump -nvi eth0

tcpdump -nvi eth0 icmp

tcpdump -nvxi eth0

tcpdump -nxi eth0

tcpdump -nxi eth0 not port ssh

other network traffic

be creative!

there's so much I hadn't been listening into, like

• online games - I guess a wide variety is waiting here, and some will be very distinctive
• doomscrolling on the various platforms
• network backup
• IoT stuff
• remote desktop protocols
• write your own code
• bundle together ping, netcat, socat, nmap and whatnot and make them MIDI controllable through a software registered MIDI client

if possible disable encryption, then you can profit from not only timing pattern (of white noise), but also listen in on the plaintext payload. the NSA loves plaintext.

the tech bits

the NSA selector is a fast ethernet (FE=100Mbps) network switch with three ports. the two front ports are switched, and the third port is only available internally as 4 bit MII bus. it is configured as mirror port of the two front ports and wired to a 4 bit DAC and a low-pass-filter.

• 4 HP wide
• current consumption:
  • 12V : 100mA
  • 5V : nothing
  • -12V : 2mA

assembling the kit

both versions are available in my store, a fully assembled eurorack module and a kit version where you have to solder on the front plate.

watch my NSA selector kit assembly video on youtube.

We will try to ship in-stock products within two to three business days.

For international customers outside the EU: Import duties, taxes and all other fees are the recipient's responsibility.