Glossary
Real-Time Factor (RTF) measures how fast an ASR system processes audio relative to the audio's duration. RTF = 1.0 means the system processes 1 hour of audio in 1 hour ("real time"). RTF = 0.1 means 1 hour of audio in 6 minutes (10× faster than real time). Lower is better for batch processing; for streaming, RTF must stay below 1.0 or your system falls behind.
How RTF is calculated
RTF = processing_wall_time / audio_duration. If you transcribe a 30-minute file in 3 minutes, RTF = 3/30 = 0.1. RTF can be measured per-system (single GPU, single thread) or end-to-end (your entire pipeline including upload, queueing, post-processing). The latter is what users experience.
Typical RTF numbers
Whisper-1 on OpenAI's API: ~0.05-0.15 RTF (fast GPUs, optimized inference). Whisper Large v3 on a consumer RTX 4090: ~0.1-0.2 RTF. distil-whisper on CPU: ~0.5-2.0 RTF (CPU-only inference is slow). Streaming ASR systems: must stay under 1.0 — typically 0.3-0.7 with end-pointing. Old hybrid HMM systems on a fast CPU: ~0.5-1.5.
Why RTF matters less than you think
For batch transcription, RTF is dominated by network round-trip time (uploading the file) for short audio, and by the actual ASR for long audio. Our pipeline for a 5-minute file is RTF ~0.4 because upload+queueing+post-processing dominates the ~10 seconds Whisper actually spends. For a 60-minute file, RTF drops to ~0.05 because Whisper's processing scales linearly with audio length while overhead is constant. Headline RTF marketing numbers should be taken with a grain of salt.
In practice
Upload a 60-minute podcast. Network upload at typical home internet: 30 seconds. Queueing (busy server, waiting for a worker): 5 seconds. Whisper processing: 90 seconds. Post-processing (formatting, JSON parsing, DOCX export): 10 seconds. Total: 135 seconds = 2.25 minutes. End-to-end RTF = 2.25 / 60 = 0.0375 — about 27× real-time.
Related terms
Try fast transcription
Get started — freeLast updated: September 20, 2026
Frequently Asked Questions
What does RTF (Real-Time Factor) mean in transcription?
How fast an ASR system processes audio relative to its duration. RTF = 0.5 means processing 1 hour of audio takes 30 minutes. Whisper-1 on a fast GPU runs at RTF ~0.1-0.3.
Why does RTF (Real-Time Factor) matter when choosing a transcription workflow?
RTF (Real-Time Factor) affects how you understand transcript quality, timing, compatibility, or the technology behind speech-to-text results.
Where can I apply RTF (Real-Time Factor) on Transcript.you?
Use the related workflow at /ai-transcription when you want to see how this glossary concept connects to an actual transcription task.
Browse the full transcription glossary or read the complete guide to AI transcription.