Speech-to-Text Infographics

How Fast Can Recorded Audio Become Text?

A 42.1-minute recording, 33.9 seconds of execution. See the measured averages across six input durations.

Speech is Cheap

42 Minutes In.
34 Seconds To Process.

Mean audio duration and execution time for 30⁠–⁠60 min recordings.

42.1 min
33.9 sec

mean audio duration

mean execution time

Input durationMean audioMean executionJobs
Up to 1 min27.8 sec3.7 sec28,900+
1–5 min2.3 min5.7 sec15,300+
5–15 min8.9 min11.1 sec7200+
15–30 min21.8 min20.6 sec4100+
1–2 hours79.3 min68.8 sec1300+
74.5×

audio duration divided by execution time
for the highlighted bucket

Live speech arrives at 1× speed. Speech is Cheap focuses on prerecorded audio, where the whole recording is ready to process faster than playback.

Source: Speech is Cheap Bench, production timing snapshot 2026-09-15 14:19 UTC. Latest benchmark group, completed jobs since 2026-07-15; mixed settings. Execution excludes queue delay. RTFx = total audio duration ÷ total execution time. Job-count lower bounds checked 2026-09-15 20:59 UTC. Bucket lower bounds exclusive, upper bounds inclusive. Observed means, not guarantees.

Read the Timing Behind the Headline

The highlighted group contains recordings longer than 30 minutes and up to 60 minutes. Its 74.5× ratio divides total audio duration by total execution time across the same completed jobs. Execution excludes queue delay; it is not the full time from sending a request to receiving a result. These are observed averages across mixed request settings, not a latency guarantee.

Put Your Recordings to Work

Review included minutes, overage and optional features. Choose the Speech is Cheap API plan that fits your recording workflow.