How Voice Is Digitized and Transmitted Over IP

Turning Sound Into Something a Network Can Carry

A traditional phone call travels as a continuous analog electrical signal down a dedicated circuit-switched line (covered earlier in this project). VoIP (Voice over IP) instead converts speech into digital data and sends it as packets over an ordinary IP network - the same kind of network carrying web traffic and everything else covered throughout this project.

The Digitization Process, Step by Step

1. Microphone captures analog sound waves
2. Sampling: the analog signal is measured at very frequent, regular intervals
3. Quantization: each sampled measurement is converted into a digital number
4. Compression (via a codec): reducing the amount of data needed per second
5. Packetization: small chunks of this digital audio are wrapped into IP packets and sent

Codecs: Balancing Quality Against Bandwidth

A codec is the specific algorithm used to compress and decompress the digitized voice data. Different codecs offer different trade-offs between audio quality and how much bandwidth they require - a call over a limited connection might deliberately use a codec that sacrifices some quality to use noticeably less bandwidth.

Why VoIP Almost Always Uses UDP, Not TCP

As covered in the TCP vs UDP note earlier in this project, TCP's retransmission of lost data adds delay - genuinely undesirable for a live conversation, where a late, out-of-date piece of audio is often worse than simply skipping it entirely. This is exactly why VoIP overwhelmingly uses UDP rather than TCP.

Turning Packets Back Into Sound at the Other End

The receiving device reverses this entire process: reassembling the arriving packets, decompressing the audio data, and converting the digital numbers back into an analog signal that drives a speaker - ideally happening quickly enough that the conversation feels natural and close to real-time.