CekuraBench

GPT-4o mini TTS

Every voiceCompare

7th of 26 voices at sounding natural

Naturalness (1 to 5): 4.43

GPT-4o mini TTS: 7th of 26 voices at sounding natural. Its weakest is medical terms, 23rd of 26.

Last updated Methodology by Dileep Chagam

GPT-4o mini TTS at a glance

Where it ranks among all 26 voices, in plain words.

Good at

Nothing in the top quarter of the voices.

Not so good at
  • Misreads medical terms23rd of 26
  • Slow to start talking21st of 26

Hear GPT-4o mini TTS

The first read of each test sentence, unedited.

GPT-4o mini TTS

GPT-4o mini TTS head to head

Pick any voice to set against it. Every pair has its own page.

Compare
MeasureGPT-4o mini TTSElevenLabs v4
  1. Non-streaming688 ms678 ms
  2. StreamingCan't stream561 ms
  3. StoppingNo stop messageFinishes first
  4. Accuracy82.7%80.4%
  5. Naturalness4.434.42
  6. Price / 1M$25.70$80.00
  7. Measures won3 of 61 of 6
The check marks the better value on each measure. Every pair has its own page with the full numbers.Open this pair · 325 pairs in all

Its closest rivals on speed

GPT-4o mini TTS FAQ

  • GPT-4o mini TTS starts in 688 ms on a typical sentence (median) and 1.50 s on the slowest 1 in 10 (p90), 21st of 26 voices.

  • No. GPT-4o mini TTS needs the whole reply before it starts.

  • GPT-4o mini TTS has no stop message, so the agent has to close the connection to stop it.

  • GPT-4o mini TTS read 82.7% of the hard parts right, as speech-to-text heard it. Its weakest kind is codes and addresses, at 77.5%. This is an estimate, not a listening test.

  • GPT-4o mini TTS lists at $25.70 per million characters.