Skip to main content
← Back to glossary
Voice infrastructure

Call transcription

What is Call transcription?

Call transcription is the automatic conversion of a phone call's spoken audio into readable text — real-time, as speech streams during a live call, or post-call, from a finished recording. It runs on automatic speech recognition, and the resulting transcripts power search across call history, agent coaching and quality assurance, compliance monitoring, call summaries, and downstream AI workflows. Accuracy varies with audio quality, background noise, and how specialized the vocabulary is, so most platforms keep the recording alongside the transcript.

More detail

Real-time transcription streams text while the call is still in progress, which is what lets an AI voice agent respond to what was just said, agents-assist hints surface mid-call, or a keyword alert fire before the caller hangs up; post-call transcription from a recording trades that latency for a cleaner, full-context transcript used by quality assurance and compliance review.

Unlike voicemail transcription, which handles a short single-speaker message left after a call, call transcription runs over a long two-way conversation — most platforms diarize the two call legs, so the transcript reads as caller versus agent instead of one undifferentiated block of text.

Frequently asked

What is the difference between call transcription and voicemail transcription?
Voicemail transcription converts a short, one-way message recorded after a call is missed; call transcription converts the live two-way conversation itself — either in real time while the call runs or afterwards from the recording — and separates the speakers so a reader can tell caller from agent.
Is real-time or post-call transcription more accurate?
Post-call transcription from the recording is usually slightly more accurate because the model can weigh the full context before committing to words, while real-time transcription prioritizes low latency so text and alerts appear while the call is still in progress. In both cases accuracy depends on audio quality, background noise, and how uncommon the spoken vocabulary is.

Build it on Orbit

Voice, messaging, email, video, and AI agents on one platform and one pay-as-you-go bill. Start free — no credit card required.