Speech benchmarks

Plain sentence

Why we chose this prompt: A short everyday sentence with a proper noun and two spoken-out times. It is the baseline: if a model drops, adds or changes a word here, nothing else it does matters.

Prompt

Route 6 Espresso is open till nine, and the last train leaves at eleven fifty-six.

Sort by

tara
Cost
--
Generation time
6.3s
Correctness
100%
provider default
Cost
$0.001
Generation time
0.7s
Correctness
100%
provider default
Cost
$0.001
Generation time
1.2s
Correctness
100%
provider default
Cost
$0.001
Generation time
0.9s
Correctness
100%
Zephyr
Cost
--
Generation time
3.6s
Correctness
100%
af_alloy
Cost
$0.000051
Generation time
0.5s
Correctness
100%
en-US-Harper:MAI-Voice-2
Cost
--
Generation time
1.2s
Correctness
100%
en-US-Harper:MAI-Voice-2
Cost
$0.001
Generation time
1.9s
Correctness
100%
English_expressive_narrator
Cost
$0.008
Generation time
1.2s
Correctness
100%
English_expressive_narrator
Cost
--
Generation time
1.0s
Correctness
100%
loongjohn
Cost
$0.001
Generation time
2.0s
Correctness
100%
eve
Cost
$0.001
Generation time
1.1s
Correctness
100%
conversational_a
Cost
--
Generation time
67.5s
Correctness
95%
en_paul_sad
Cost
$0.000
Generation time
5.0s
Correctness
89%
longanlingxin
Cost
$0.002
Generation time
1.9s
Correctness
80%
aura-2-thalia-en
Cost
$0.002
Generation time
1.5s
Correctness
76%

All benchmarks

Media benchmarks

Compare generation quality across media models