Converting binary to base64 makes the data 33% bigger, you should not do that when you have limited bandwidth
this post was submitted on 12 Nov 2025
8 points (100.0% liked)
Meshtastic
1426 readers
19 users here now
A community to discuss Meshtastic (https://meshtastic.org/docs/introduction)
founded 2 years ago
MODERATORS
So, hex encoding of the aes cipher made it bigger. base 85 is slightly smaller than 64. The goal was to make it copy paste safe I think that's the best option
There is no way to make this work. As others told you, hex made it bigger, because hex is like equivalent to base16. Every digit in base64 is 6 bit, while in hex it's 4 bit. Raw bytes would be equivalent to base256.
You will not get a good result with this technology, it just can't handle the bandwidth, and will make it worse for everyone else by consuming their shared bandwidth.
Instead, try to look into midi and sound fonts, that might give you a more sustainable solution.
It definitely works. I tested a five second audio clip at 2kbs. It's about 15 manual copy pastes at 200 character limit...but it works.
This isn't going to be used it's a very manual process...but I'm glad I can do it!
Ok, let me be more specific. Of course it delivers the audio, so you can technically say that i "works", that wasn't up to debate, you can deliver anything, actually. The problem is that it will never be viable. You shouldn't abuse the network, as it is a shared channel, when you use 2kbps to send your audio, no one else's can use it for messages, and viceversa, relaying many messages will limit your bandwidth.
You are trying to fit a square peg in a round hole, while instead there are tools designed for this purpose. It's kind of silly developing a roundabout way to send audio through meshtastic while every other amateur radio ecosystem is already designed for audio.
Good point, I guess using hexadecimal would shrink the data even further than using base64.
The compression was implemented to reduce the audio input drastically. It can compress audio down to 6kb/s and you can still interpret voice clearly.
I should try and optimize even further to shrink the final payload.