Everything worth knowing, nothing that isn't.
How a programmable voice API works
Your application makes a REST call (or answers an inbound call) and the platform sets up the call over SIP to the PSTN and mobile networks. As the call progresses — ringing, answered, digit pressed, hung up — the platform posts webhooks back to your app, and you respond with instructions to play audio, gather input, record, or bridge. It's the voice equivalent of an SMS API. See /products/voice and /products/voice.
The core building blocks
DID numbers to receive inbound calls; IVR to play prompts and gather DTMF or speech; text-to-speech and speech recognition for dynamic audio; call recording and transcription; and conferencing/bridging to connect callers to agents. Each is exposed as an API primitive you compose into a flow. See /answers/what-is-did and /answers/what-is-ivr.
Voice OTP and its trade-offs
Voice OTP places an automated call that reads a one-time code aloud — useful as a fallback when SMS OTP fails, for landlines, or where a market is being SMS-pumped. It reaches numbers SMS can't and is harder to intercept silently, but costs more per attempt and takes longer to complete than an SMS. Robust verification fails over between SMS, voice, and flash call. See /answers/what-is-otp and /answers/what-is-flash-call-verification.
The metrics that matter: ASR, ACD, and PDD
Answer-Seizure Ratio (ASR) measures the share of call attempts that connect; Average Call Duration (ACD) measures how long answered calls last; and Post-Dial Delay (PDD) measures how long callers wait before ringing. Together they reveal route quality — low ASR or high PDD signals a poor carrier path. Size trunks and agents with the /tools/erlang-calculator. See /answers/what-is-asr, /answers/what-is-acd, and /answers/what-is-pdd.
Benefits of building on a voice API
No telephony hardware or carrier contracts to manage, global reach through one interconnect, pay-per-use billing, and the ability to change call flows in code rather than reprogramming a PBX. It puts inbound and outbound voice on the same developer footing as SMS and WhatsApp. See /products/voice.
Limitations and things to plan for
Voice is rate-limited by calls per second (CPS) and concurrent-channel capacity, so bursts must be paced; regulations (DNC lists, caller-ID/CLI rules, STIR/SHAKEN in North America) constrain outbound dialing; and per-minute economics differ sharply from per-message SMS. Model concurrency and CPS before a large campaign. See /answers/what-is-cps and /answers/what-is-dnc.