bits is bits

In a nutshell, copper Ethernet sends 2.5V pulses representing ones and zeroes over twisted-pair cables. 

One wire of the pair carries +2.5V pulses

The other wire carries -2.5V pulses

The pulses are identical in pattern but opposite in voltage, so that if the carrier voltage were altered, say by EMI, the difference between opposite pulses would be unchanged at 5V peak to peak, allowing the network interface to correctly differentiate ones from zeroes regardless of how dirty the signal may be. 

That’s the physical, low-level layer. On top of that come advanced protocols such as TCP with their own error-correction protocols.

Note that none of this applies to fiber. Fiber sends pulses of light through a glass medium where light=1 and dark=0, and these light pulses are completely impervious to EMI.

Here is a very good article on twisted-pair Ethernet. Imo it is required reading for audiophiles who have Ethernet components in their systems - in other words, anyone who streams music - and who want to understand which factors can and can’t affect the sound quality of their systems.

https://www.practicalnetworking.net/stand-alone/ethernet-wiring/

I replaced the ethernet cable between my streamer and switch with SFP optical fibre. As far as I can discern, it made absolutely zero difference to sound quality.

 

That’s in a highly resolving system. Both the Linn Klimax DSM/3 streamer and Melco S100 switch have SFP ports and I used the often recommended Finisar optical modules. 

 

Other changes I have made in the past like adding the Melco switch and upgrading the streamer have been beneficial, but the SFP optical fibre wasn’t. Others using the same streamer have reported improvement from changing to fibre. Maybe the Melco is already providing sufficient isolation.

@newton_john that’s most likely the case. A person would expect the multi-thousand dollar Melco switch to have a completely noise-free output, and therefore the fiber run downstream of the switch would become redundant in terms of noise reduction and yield no sonic benefit.

In regard to network altering bits, https://www.google.com/search?client=firefox-b-1-d&q=can+network+alter+bits&udm=50&fbs=AIIjpHxU7SXXniUZfeShr2fp4giZ1Y6MJ25_tmWITc7uy4KIeoJTKjrFjVxydQWqI2NcOhZVmrJB8DQUK5IzxA2fZbQF4YL5sNSRJGgx0e9Z9AxExzjE4_ynshmXB4KOs3cwRUeWWsKhEolaqKblYfDXcX8o44lrMqH67kppK0pOExd7q2yJDRQQKlpEBHmw6_MVnVB63-D4QfZ_FNhtd2-sQaouA4jq5w&ved=2ahUKEwiZ6cTNlsiQAxVBnokEHf3VMcMQ0NsOegQIAxAB&aep=10&ntc=1&mstk=AUtExfDDuUPdWFfnGdINKi92dM-0CQQv45KQzyN3TxSAjuzkjTDZj4cKP_6bIpma1AUzx6NavD42tl140w4xi3Sgkid8-P-eSrD5v4x2PcZaclIqGC-OhQgO0V6i3YnUcz_miFgRRBn8eDnB6u25ZN4KfXbSRDu5G-M2odA&csuir=1. 

Now perhaps AI full of it.

Now perhaps AI full of it.

@sns 

Yes.

From a 30 year network engineer, you know, an open-minded one who actually listens to his gear: 

"Sure, network protocols prevent dropped bits, so the music data itself stays complete. But the extra activity inside the streamer can subtly compress the soundstage and reduce the microdynamics that give recordings their sense of space and life."

https://www.headphonesty.com/2025/09/tidal-qobuz-connect-ruining-hifi-sound-engineer/