What is Speech-to-Text (STT)?
Speech-to-Text: Technology that converts spoken audio into written text, also called automatic speech recognition (ASR), and the first step in most voice agents.
Speech-to-text, or automatic speech recognition, turns audio into a transcript. For voice agents it runs in streaming mode: audio is sent in small chunks and partial transcripts come back while the caller is still speaking, so the system can start working before they finish. Providers include Deepgram, AssemblyAI, Speechmatics, Google, and OpenAI, plus open models like Whisper that can be self-hosted.
Accuracy is usually reported as word error rate, but for business calls what matters is whether the important words come through: names, phone numbers, dates, product names, and addresses. Phone audio is compressed and noisy, which makes this harder than transcribing a podcast. Many providers let you boost custom vocabulary so your brand names and product terms are recognized.
Why it matters when you build or buy AI
A voice agent cannot do better than its transcript. Test STT on recordings of your actual callers, with your accents, background noise, and vocabulary, not on a vendor demo. Pay special attention to how numbers and spellings are captured, and design the agent to read back critical details like phone numbers and dates before acting on them.
Example
A plumbing company adds its service area town names and common part names to the STT custom vocabulary after test calls show addresses being transcribed wrong.
Related terms
More in Voice AI
Book 20 minutes, or send a brief
A former CTO replies. You leave with a written plan and a fixed estimate.
Prefer LinkedIn? Connect with Harish.