Benchmark Almanac

MiniMax TTS Multilingual Test Set

In use

Text-to-speech models clone two Common Voice speakers per language and read 100 test sentences in each of 24 languages. Scored by word error rate and speaker similarity.

Released with MiniMax-Speech. MiniMax plans to add languages, so the set may grow. Lower error rate is better.

Made by
MiniMax
Released
May 2025
Kind
Speech recognition and translation
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
Many languages
In and out
Text, Audio → Audio
Interaction
Single turn
Items from
Another benchmark
Refresh
One fixed set
Run by
Anyone
Last in a launch post
September 2026
Last model added
September 2026
1
Model scored in lab cards
1
Launch post lists it, from 1 lab
2
System cards report it, from 1 lab

Ming-3.0-tts-0.5B leads with 0.36%

Best score per model as labs report them in their own cards, drawn striped. Labs pick their own settings, so these compare less cleanly than independent runs.

  1. 1 Ming-3.0-tts-0.5B 0.36%

Source: Labs’ model cards

#ModelScoreSettingDateSource
1 Ming-3.0-tts-0.5B 0.36% Sep 2026 Lab model card

Scores over time

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

Reported by the lab in its model card
0% 20% 40% 60% 80% 100% Ming-3.0-tts-0.5B 0% 20% 40% 60% 80% 100% Ming-3.0-tts-0.5B

Source: Labs’ model cards

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Ant Group
DateLaunch postScoreAs the post writes it
Sep 22, 2026 Ming-3.0-tts-0.5B 0.365 German WER ↓

2 system cards report it, from 1 lab

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Apr 17, 2026 Qwen3.5-Omni-Plus technical report TTS multilingual test set, WER per language, no overall (Table 9): zh 0.695, en 0.631, de 0.447, it 0.503, pt 1.221, es 0.862, ja 3.479, ko 1.458, fr 2.430, ru 3.182, th 2.170, id 0.823, ar 2.602, vi 1.143, tr 0.938, fi 2.784, pl 1.427, hi 6.444, nl 1.238, cs 2.929. Speaker similarity per language in Table 9, custom-voice WER in Table 12 Capabilities
Sep 22, 2025 Qwen3-Omni-30B-A3B technical report MiniMax Multilingual Test Set, 10 languages, content consistency WER from 0.716 (Chinese) to 3.986 (Russian) and speaker similarity 0.689 to 0.778, no overall score (Table 14) Capabilities

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 4 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 pass 4 notes
  • Note Independent runs Only labs’ own numbers, so no one else has checked them. Scores from launch posts (1 score).
  • Note Models scored Only 1 model we track.
  • Pass Newest score The newest is from September 2026, for Ming-3.0-tts-0.5B.
  • Note Frontier models None of its models are on the capabilities index.
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 1 note
  • Note Importer No public leaderboard to read.
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