Text-to-speech · measured leaderboard

Only rows measured by CodeSOTA rank here.

The May hard-text rows are withheld: their manifests contain placeholder hashes and do not account for the claimed 30 audio samples. No auditable measured hard-text ranking is available until complete artifacts are restored. Blind Elo is a separate preference study and does not supply hard-text accuracy or latency results.

Open separate blind Elo study Open reported registry Open watchlist Polish track

Separate study · not benchmark-ranked

Active blind Elo sample pool

These seven male-voice systems have the shared prompt audio ready for preference voting. They will not enter the hard-text measured ranking until ASR transcripts, diffs, entity scoring, latency logs, and artifacts are published for that benchmark.

Vote on blind Elo

Model	Vendor	Voice condition	Elo clips	Measured benchmark status
Gradium TTS gradium-tts:kent	Gradium	Kent	30	audio ready · scoring pending
Gradium TTS gradium-tts:damon	Gradium	Damon	30	audio ready · scoring pending
Gradium TTS gradium-tts:russell	Gradium	Russell	30	audio ready · scoring pending
Kokoro v1.0 hexgrad/kokoro-82m:am_michael	Hexgrad	am_michael	30	audio ready · scoring pending
Speech-02 Turbo minimax/speech-02-turbo:english-deep-voiced-gentleman	MiniMax	English_Deep-VoicedGentleman	30	audio ready · scoring pending
Speech-02 HD minimax/speech-02-hd:english-deep-voiced-gentleman	MiniMax	English_Deep-VoicedGentleman	30	audio ready · scoring pending
Qwen3 TTS qwen/qwen3-tts:aiden	Qwen	Aiden	30	audio ready · scoring pending
Chatterbox Turbo resemble-ai/chatterbox-turbo:andy	Resemble AI	Andy	30	audio ready · scoring pending
Chatterbox Turbo resemble-ai/chatterbox-turbo	Resemble AI	default study voice	30	audio ready · scoring pending
ElevenLabs v3 elevenlabs/v3:james	ElevenLabs	James	30	audio ready · scoring pending
XTTS v2 coqui/xtts-v2:damien-black	Coqui	Damien Black	30	audio ready · scoring pending

Rank	Model	Benchmark	Verification	Entity acc.	WER	CER	p95 TTFB	CI	Artifacts
No verified hard-text benchmark rows are currently published. Prior placeholder-backed rows are withheld pending complete audio manifests, real hashes and sample-level outputs.

Harness commands

codesota-tts synth --model <id> --eval <track> --out runs/<run_id>
codesota-tts score --run runs/<run_id> --metrics wer,cer,entity,utmos,latency
codesota-tts report --run runs/<run_id> --publish

TTS Eval v2 tracks

clean-read-en

Harvard-style clean sentences

UTMOS, WER, CER, latency

hardtext-en

Numbers, dates, currencies, addresses, acronyms, emails, URLs, product codes

WER, CER, critical entity accuracy, severe errors

hardtext-pl

Polish diacritics, dates, currencies, addresses, abbreviations

Polish CER, entity exactness, abbreviation handling

longform

5-15 minute narration/dialogue

voice drift, omission/repetition rate, long-run WER

cloning

Speaker preservation with reference audio

speaker similarity, WER

controllability

Emotion, speed, pitch, whisper, pauses, delivery style

control adherence and acoustic-channel movement