A cross-platform softphone built by a Bulgarian team. It only does the client, which keeps it neutral towards every SIP provider.
# The same 'call connected' takes a completely different path on every system. Cross-platform is not compiling six times; it is maintaining six sets of compatibility logic.
To an outsider cross-platform means compiling one codebase for several systems. Anyone who has worked with audio knows the real cost is that every system manages audio devices differently.
Windows has exclusive mode, hot-plug events and a zoo of virtual sound cards. macOS has permission prompts and aggregate devices. Linux has the divergence between PulseAudio and PipeWire. The same capture code behaves differently on two systems.
Harder still is the default device question: the user plugs in a headset, the system switches its default, and an established call does not automatically follow. Whether it should, and when, is a product decision, not a technical detail.
iOS audio sessions must declare their purpose explicitly; declare it wrongly and audio comes out of the earpiece or goes silent after connecting. Android fragmentation is worse: vendor battery policies, background limits and push channels all differ, so one version behaves very differently across devices.
Testing a cross-platform softphone usually costs more than building it.
Bluetooth headsets involve two profiles: hands-free for calls (narrow band, poor quality) and media playback (good quality, no capture). Switch at the wrong moment and the user hears something like a cheap radio, or discovers the other side cannot hear them.
Because users will not forgive you for the difficulty of cross-platform. They remember that this app connects with no sound. In communications products, that kind of basic failure erases every feature advantage at once.