Left is faster. Higher reads more hard parts right.
GPT-4o mini TTS is better on naturalness. Grok TTS is better on non-streaming latency, streaming latency, stopping, accuracy and price.
Last updated Methodology by Dileep Chagam
Measure by measure, the same sentences, the same place.
An estimate: speech-to-text transcribers scored it, not people.
The same sentence, read by each voice.
Grok TTS. GPT-4o mini TTS: 688 ms. Grok TTS: 358 ms.
Grok TTS. GPT-4o mini TTS: 688 ms. Grok TTS: 358 ms.
Grok TTS. GPT-4o mini TTS: Can't stream. Grok TTS: 157 ms.
Grok TTS. GPT-4o mini TTS: No stop message. Grok TTS: 0 ms.
Grok TTS. GPT-4o mini TTS: 82.7%. Grok TTS: 87.4%.
Grok TTS. GPT-4o mini TTS: $25.70. Grok TTS: $15.00.