Left is faster. Higher reads more hard parts right.
ElevenLabs v3 is better on non-streaming latency, accuracy and price. ElevenLabs v4 is better on streaming latency and naturalness.
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.
ElevenLabs v3. ElevenLabs v3: 320 ms. ElevenLabs v4: 678 ms.
ElevenLabs v3. ElevenLabs v3: 320 ms. ElevenLabs v4: 678 ms.
ElevenLabs v4. ElevenLabs v3: Can't stream. ElevenLabs v4: 561 ms.
Neither has a figure. ElevenLabs v3: No stop message. ElevenLabs v4: Finishes first.
ElevenLabs v3. ElevenLabs v3: 87.2%. ElevenLabs v4: 80.4%.
ElevenLabs v3. ElevenLabs v3: $40.00. ElevenLabs v4: $80.00.