Benchmark Almanac

Real World VoiceEQ

Gemini 3.8 Flash 0.92 leads since September 2026
Not in model cards

Voice systems recognize, produce and respond to acoustic information across speech recognition, speech to speech, speech understanding and text to speech tasks.

Hume evaluates more than 40 voice models across 15 or more dimensions and 60 or more metrics, using more than one million human ratings. The benchmark covers four modalities and publishes public leaderboards.

Made by
Hume AI
Released
July 2026
Kind
Audio understanding and voice
Level
Expert
Scored
Graded by people
Answers
Open-ended
Tasks
Public, plus a private test set
Measures
Preference
In and out
Text, Audio → Text, Audio
Interaction
Single turn
Run by
A host
Strongest model tested
Gemini 3.8 Flash #24 on the capability index
Last model added
September 2026
Scores refreshed
November 21st, 2026
39
Entries on its own leaderboard, 5 models we track

Gemini 3.8 Flash leads with 0.92 among models released by September 2026 Gemini 3.8 Flash leads with 0.92 among models released by September 2026

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. This view shows only models we track, each at its best.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemini 3.8 Flash 0.92
  2. 2 elevenlabs/eleven_v4_turbo 0.91
  3. 3 google/gemini-3.8-flash-lite 0.91
  4. 4 elevenlabs/eleven_v4 0.90
  5. 5 Gemini 2.5 Pro Jun 2025 0.88
  6. 6 Gemini 2.5 Flash Sep 2025 0.86
  7. 7 cartesia/sonic-3.6 0.84
  8. 8 Grok TTS 0.79
  9. 9 google/gemini-3.1-flash 0.78
  10. 10 elevenlabs/eleven_v3_conversational 0.77
  11. 11 cartesia/sonic-3.5 0.76
  12. 12 speechify/simba-3.2 0.75
  13. 13 GPT-4o Mini TTS 0.74
  14. 14 fishaudio/s2-pro 0.72
  15. 15 elevenlabs/eleven_v3 0.71
  16. 16 openai/tts-1 0.69
  17. 17 google/chirp3-hd 0.69
  18. 18 openai/tts-1-hd 0.69
  19. 19 inworld/tts-2-flash 0.68
  20. 20 inworld/tts-1-max 0.65
  21. 21 elevenlabs/eleven_multilingual_v2 0.64
  22. 22 deepgram/flux 0.61
  23. 23 deepgram/aura-2 0.59
  24. 24 smallest/lightning_v3.1 0.58
  25. 25 inworld/tts-2 0.58
  26. 26 elevenlabs/eleven_flash_v2_5 0.57
  27. 27 microsoft/VibeVoice-1.5B 0.56
  28. 28 hexgrad/Kokoro-82M 0.54
  29. 29 inworld/tts-1 0.54
  30. 30 resemble-ai/chatterbox 0.51
  31. 31 qwen/Qwen3-TTS-12Hz-1.7B 0.49
  32. 32 hume/tada-3b-ml 0.48
  33. 33 sesame/csm-1b 0.38
  34. 34 coqui/xtts-v2 0.37
  35. 35 bosonai/higgs-audio-v2 0.35
  36. 36 bosonai/higgs-audio-v3-tts-4b 0.29
  37. 37 indexteam/indextts-2 0.18
  38. 38 nari-labs/dia 0.17
  39. 39 parler-tts/parler-tts 0.04

Source: Official Real World VoiceEQ leaderboard

#EntryScoreSettingDateSource
1 Gemini 3.8 Flash 0.92 Text-to-Speech Sep 2026 Official leaderboard
2 elevenlabs/eleven_v4_turbo 0.91 Text-to-Speech Official leaderboard
3 google/gemini-3.8-flash-lite 0.91 Text-to-Speech Official leaderboard
4 elevenlabs/eleven_v4 0.90 Text-to-Speech Official leaderboard
5 Gemini 2.5 Pro Jun 2025 0.88 Text-to-Speech Jun 2025 Official leaderboard
6 Gemini 2.5 Flash Sep 2025 0.86 Text-to-Speech Sep 2025 Official leaderboard
7 cartesia/sonic-3.6 0.84 Text-to-Speech Official leaderboard
8 Grok TTS 0.79 Text-to-Speech Apr 2026 Official leaderboard
9 google/gemini-3.1-flash 0.78 Text-to-Speech Official leaderboard
10 elevenlabs/eleven_v3_conversational 0.77 Text-to-Speech Official leaderboard
11 cartesia/sonic-3.5 0.76 Text-to-Speech Official leaderboard
12 speechify/simba-3.2 0.75 Text-to-Speech Official leaderboard
13 GPT-4o Mini TTS 0.74 Text-to-Speech Mar 2025 Official leaderboard
14 fishaudio/s2-pro 0.72 Text-to-Speech Official leaderboard
15 elevenlabs/eleven_v3 0.71 Text-to-Speech Official leaderboard
16 openai/tts-1 0.69 Text-to-Speech Official leaderboard
17 google/chirp3-hd 0.69 Text-to-Speech Official leaderboard
18 openai/tts-1-hd 0.69 Text-to-Speech Official leaderboard
19 inworld/tts-2-flash 0.68 Text-to-Speech Official leaderboard
20 inworld/tts-1-max 0.65 Text-to-Speech Official leaderboard
21 elevenlabs/eleven_multilingual_v2 0.64 Text-to-Speech Official leaderboard
22 deepgram/flux 0.61 Text-to-Speech Official leaderboard
23 deepgram/aura-2 0.59 Text-to-Speech Official leaderboard
24 smallest/lightning_v3.1 0.58 Text-to-Speech Official leaderboard
25 inworld/tts-2 0.58 Text-to-Speech Official leaderboard
26 elevenlabs/eleven_flash_v2_5 0.57 Text-to-Speech Official leaderboard
27 microsoft/VibeVoice-1.5B 0.56 Text-to-Speech Official leaderboard
28 hexgrad/Kokoro-82M 0.54 Text-to-Speech Official leaderboard
29 inworld/tts-1 0.54 Text-to-Speech Official leaderboard
30 resemble-ai/chatterbox 0.51 Text-to-Speech Official leaderboard
31 qwen/Qwen3-TTS-12Hz-1.7B 0.49 Text-to-Speech Official leaderboard
32 hume/tada-3b-ml 0.48 Text-to-Speech Official leaderboard
33 sesame/csm-1b 0.38 Text-to-Speech Official leaderboard
34 coqui/xtts-v2 0.37 Text-to-Speech Official leaderboard
35 bosonai/higgs-audio-v2 0.35 Text-to-Speech Official leaderboard
36 bosonai/higgs-audio-v3-tts-4b 0.29 Text-to-Speech Official leaderboard
37 indexteam/indextts-2 0.18 Text-to-Speech Official leaderboard
38 nari-labs/dia 0.17 Text-to-Speech Official leaderboard
39 parler-tts/parler-tts 0.04 Text-to-Speech Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 3.8 Flash 0.92 Text-to-Speech Sep 2026 Official leaderboard
2 Gemini 2.5 Pro Jun 2025 0.88 Text-to-Speech Jun 2025 Official leaderboard
3 Gemini 2.5 Flash Sep 2025 0.86 Text-to-Speech Sep 2025 Official leaderboard
4 Grok TTS 0.79 Text-to-Speech Apr 2026 Official leaderboard
5 GPT-4o Mini TTS 0.74 Text-to-Speech Mar 2025 Official leaderboard

Rankings by part

We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Acting / Role-fit, Expressiveness and Voice identity. Gemini 2.5 Pro leads 3 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Acting / Role-fit

Mean human rating from the Text-to-Speech panel. 5 models we track have a score here.

  1. 1 Gemini 3.8 Flash 3.69
  2. 2 Gemini 2.5 Flash Sep 2025 ↑1 3.64
  3. 3 Gemini 2.5 Pro Jun 2025 ↓1 3.63
  4. 4 GPT-4o Mini TTS ↑1 2.88
  5. 5 Grok TTS ↓1 2.78

2 models have held number one on the official leaderboard since its release in July 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Own leaderboard Not tested since Sep 2026

Source: Official Real World VoiceEQ leaderboard

Scores over time

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

Plot with other benchmarks
On the benchmark’s own leaderboard
0.70 0.80 0.90 2025 2026 0.70 0.80 0.90 2025 2026

Source: Official Real World VoiceEQ 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 (39 entries).
  • Pass Models scored 5 models we track, plus 34 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: 5 rows for models we track and 34 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