June 2026 Summaries

Miravoice conducted a comprehensive study to examine the impact of AI voice selection on user engagement during phone surveys, presenting their findings at the American Association for Public Opinion Research conference. The study involved nearly 100,000 calls made to landlines using 12 different AI voices from providers Rime, ElevenLabs, and Google, assessing metrics such as hang-up rates, cooperation rates, and do-not-call rates. Results showed Rime's voices had the lowest intro hang-up rates and highest cooperation rates, suggesting they maintained engagement better than competitors. Interestingly, Rime also had a higher do-not-call rate, interpreted as a sign of engagement rather than rejection, as people stayed on the line long enough to opt out. The study found negligible differences based on gender-coded voices, emphasizing that voice-level variations were more significant. The findings underscore the importance of testing multiple voices for specific use cases, as the choice of voice can significantly influence engagement metrics. Miravoice plans to explore further variables such as pitch, frequency, and cadence in future research to deepen understanding of AI voice performance.