Benchmark Almanac

Voice Replication Leaderboard

openbmb/VoxCPM2 4.21 tops its leaderboard Gemini 3.8 Flash 3.53 leads the models we track
Not in model cards

Text to speech systems reproduce reference speakers from short recordings. Human ratings score speaker identity, quality and naturalness, with an objective speaker similarity check.

Hume evaluates 13 models on 25 reference voices and seven prompts. Three blind raters score each clip on same speaker, quality and naturalness, with TitaNet objective similarity.

Made by
Hume AI
Released
September 2026
Kind
Speech recognition and translation
Level
Expert
Scored
Graded by people
Answers
Open-ended
Tasks
Public, plus a private test set
Measures
Preference
Language
English
In and out
Text, Audio → Audio
Interaction
Single turn
Refresh
One fixed set
Run by
A host
In views
Audio and speech
Strongest model tested
Gemini 3.8 Flash #24 on the capability index
Last model added
September 2026
Scores refreshed
November 21st, 2026
13
Entries on its own leaderboard, 1 model we track

openbmb/VoxCPM2 tops the leaderboard with 4.21 Gemini 3.8 Flash leads with 3.53

Best score per entry from the benchmark's own leaderboard, read on November 21st, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Gemini 3.8 Flash, at 3.53. This view shows only models we track, each at its best.

  1. 1 openbmb/VoxCPM2 4.21
  2. 2 meituan-longcat/LongCat-AudioDiT-3.5B 4.06
  3. 3 fishaudio/s2-pro 4.03
  4. 4 bosonai/higgs-audio-v3 3.83
  5. 5 cartesia/sonic-3.5 3.70
  6. 6 elevenlabs/eleven-multilingual-v2 3.68
  7. 7 google/gemini-3.8-flash-lite 3.68
  8. 8 qwen/Qwen3-TTS-12Hz-1.7B 3.65
  9. 9 cartesia/sonic-3.6-beta 3.63
  10. 10 inworld/inworld-tts-2 3.62
  11. 11 Gemini 3.8 Flash 3.53
  12. 12 microsoft/VibeVoice-1.5B 3.46
  13. 13 elevenlabs/eleven-v3 2.91

Source: Official Voice Replication Leaderboard leaderboard

Source: Official Voice Replication Leaderboard leaderboard

#EntryScoreSettingDateSource
1 openbmb/VoxCPM2 4.21 Official leaderboard
2 meituan-longcat/LongCat-AudioDiT-3.5B 4.06 Official leaderboard
3 fishaudio/s2-pro 4.03 Official leaderboard
4 bosonai/higgs-audio-v3 3.83 Official leaderboard
5 cartesia/sonic-3.5 3.70 Official leaderboard
6 elevenlabs/eleven-multilingual-v2 3.68 Official leaderboard
7 google/gemini-3.8-flash-lite 3.68 Official leaderboard
8 qwen/Qwen3-TTS-12Hz-1.7B 3.65 Official leaderboard
9 cartesia/sonic-3.6-beta 3.63 Official leaderboard
10 inworld/inworld-tts-2 3.62 Official leaderboard
11 Gemini 3.8 Flash 3.53 Sep 2026 Official leaderboard
12 microsoft/VibeVoice-1.5B 3.46 Official leaderboard
13 elevenlabs/eleven-v3 2.91 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 3.8 Flash 3.53 Sep 2026 Official leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
3.53 3.53

Source: Official Voice Replication Leaderboard leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Audio and video

All 695 in Audio and video
  • Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
  • LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
  • MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
  • MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
  • FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better. Phi-4-multimodal 4.00%
  • VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (13 entries).
  • Pass Models scored 1 model we track, plus 12 other entries on its leaderboard.
  • Pass Newest score The newest is from September 2026, for Gemini 3.8 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3.8 Flash (#24).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 1 row for models we track and 12 other entries at the last read.
  • Pass Last read November 21st, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check