speech model
ElevenLabs v3
ElevenLabsThe most expressive synthetic voice available, and the best voice cloning.
ElevenLabs remains the quality benchmark for synthetic speech, particularly for emotional range and for cloning a specific person’s voice from a short sample. In hiring, the cloning capability is the interesting and dangerous part: a recruiter’s real voice can front thousands of outbound calls, which is powerful and requires explicit consent and disclosure.
Variants
3 ways to run it.
Same family, different quality/cost rung. Picking the wrong one is the most common way teams overpay.
v3
Higheleven-v3Most expressive. Best for produced audio.
Flash v2.5
Fasteleven-flash-v2-5~75ms model latency for live conversation.
Turbo v2.5
Balancedeleven-turbo-v2-5Balance of expressiveness and speed.
Fit
Two audiences, one model.
The same model is a different proposition depending on what you point it at.
Recruitment fit
The best-sounding option for candidate-facing voice — with the heaviest disclosure obligations attached.
Best for
- Outbound screening calls
- Candidate voicemail and reminders
- Multilingual job ads
Strengths
- Flash v2.5 is fast enough for a natural back-and-forth screening call
- Cloned recruiter voices make outbound feel personal at volume
- Language coverage supports multi-market hiring
Watch out for
- Cloning a real employee’s voice requires their written consent
- Several jurisdictions require candidates to be told they are speaking to a machine — build the disclosure into the script, not the footer
- Per-character pricing makes long calls expensive in a way per-minute pricing hides
The numbers
Measured, not asserted.
HumanLike evals
Run on our own harness.
- Wired into HumanLike voice
- YesOur eval
