Observed Signal · Sep 10, 2026 · Product Launch · Source: Hume AI · Impact: 3.5/5

Introducing the Hume Voice Replication Leaderboard

Executive Signal Summary

On September 10, 2026, Hume AI published the Voice Replication Leaderboard, a new component of its Real World VoiceEQ Benchmark. The leaderboard ranks 11 text-to-speech models on how accurately they reproduce a specific person's voice across 25 reference voices in standard, expressive, and accented categories. Models were rated on same-speaker similarity, quality, naturalness, and objective embedding similarity; no single model led all dimensions. OpenBMB VoxCPM2 topped same-speaker similarity, while Cartesia sonic-3.6-beta achieved the highest naturalness and Inworld TTS-2 the highest quality.

Polaris7 AgentPolaris7 Strategic Assessment
High Confidence

This leaderboard provides an independent, multi-dimensional comparison of voice replication performance across leading TTS models, helping product teams make evidence-based model choices and guiding researchers toward specific failure modes. It also reinforces that naturalness and speaker identity are separate qualities, which matters for voice cloning applications in media, customer service, and accessibility.

MARKET SIGNALS

Track Hume AI Signals & Strategic Shifts

Polaris7 connects regulatory filings, primary sources, leadership changes, and M&A activity into structured intelligence. Set up your Polaris7 workspace to monitor relevant companies in your executive briefing.

  • Verified primary sources
  • Automated briefings
  • Instant watchlist setup

Key Takeaways & Evidence Grounding

  • On September 10, 2026, Hume AI launched the Voice Replication Leaderboard as an extension of its Real World VoiceEQ Benchmark.
  • The leaderboard evaluated 11 TTS models using 25 reference voices: 5 standard, 5 expressive, and 15 accented voices (6 native English and 9 non-native accents).
  • Each generated clip was rated on same-speaker similarity (1–5), quality (1–5), naturalness (1–5), and TitaNet-based cosine similarity (0–1), with three blind paid raters per clip via the Hume Human Feedback API.
  • OpenBMB VoxCPM2 ranked first in same-speaker similarity (4.21); Cartesia sonic-3.6-beta led naturalness (4.36); Inworld TTS-2 led quality (4.61); Meituan LongCat AudioDiT-3.5B led objective similarity (0.88).
  • No model led every dimension: quality scores ranged 3.85–4.61 (average ~4.30), while same-speaker scores ranged 2.91–4.21; ElevenLabs Eleven v3 ranked last on both human same-speaker and objective similarity.

Connected Companies & Entities

“Hume AI published the Voice Replication Leaderboard as part of its Real World VoiceEQ Benchmark and conducted the human evaluations via its ...”

“ElevenLabs' Eleven v3 model ranked last on both human same-speaker rating (2.91) and objective similarity (0.55) in the leaderboard.”

Primary Source Grounding & Direct Attribution
Direct Origin Attribution
Primary Reporting: Hume AI•Published: Sep 10, 2026

Related Market Signals & Shifts

Recent verified developments and strategic activity across this market segment.

Generative AI (Voice Cloning) & Terms of ServiceApr 12, 2026

I Cloned My Voice — It's No Longer Just Mine

The author used ElevenLabs and other voice-cloning tools (Speechify, Descript) to synthesize their voice, recording about 30 minutes to obtain a high-quality clone. ElevenLabs can create convincing voice replicas from short samples (the article notes a two-minute minimum is often sufficient). The piece highlights an important legal point: ElevenLabs' terms of service grant the company broad rights to reproduce, modify, publish, create derivative works from, distribute, publicly perform and otherwise use uploaded voices and personality traits. The author warns readers to check the general terms before training AI on personal material and notes practical misuse risks (for example, fooling relatives on a phone call). The story is accompanied by a t3n Tool Time video demonstrating uses of the cloned voice.

Read assessment
Product & Platform LaunchSep 17, 2026

Introducing the Hume Voice Controllability Leaderboard

Hume AI published the Voice Controllability Leaderboard on September 17, 2026, extending its Real World VoiceEQ Benchmark. The new leaderboard evaluates 17 text-to-speech models on their ability to follow direction across voice design, voice instruction, and role fit, using human ratings as the gold standard. Results show no single model dominates every controllability dimension, with Google and ElevenLabs leading most categories but exhibiting distinct weaknesses.

Read assessment
Voice Cloning / Synthetic AudioApr 4, 2026

I Cloned My Voice — Warning About Licensing

A t3n author tested readily available voice‑cloning services (naming Speechify, Descript and Elevenlabs) by recording about 30 minutes of their voice to create a synthetic clone. The generated voice sounded convincingly similar — enough to potentially deceive relatives on a call — but also left the author with some regret. The article highlights a key legal/privacy issue: Elevenlabs' terms grant the company a broad license to reproduce, modify, publish and create derived works from submitted voices. The piece urges anyone planning to train a personal voice with these tools to read the provider terms carefully and consider potential rights they may be assigning. The story is accompanied by a t3n Tool Time video episode linked on YouTube.

Read assessment

WATCHLIST & BRIEFINGS

Receive future signals for Hume AI & competitors in your briefing

Stay ahead of material M&A deals, leadership changes, account wins, and product rollouts. Polaris7 connects primary filings, company newsrooms, and trade reporting directly into your executive briefing.

  • Regular primary-source checks
  • Structured Watchlist briefings
  • Preserves current context