Darwin TTS TTS

A Qwen3-TTS variant whose talker FFN weights are blended from the Qwen3 language model for sharper cross-lingual cloning.

0/500 Àwọn àmì-àṣírí · Ṣàfihàn fún 5,000 ni gbogbo ìgbà →

Ṣẹ̀dà fun àwọn àmì-àṣírí 5,000

Àwọn Ìṣàmúlò-ètò (Àwọn Àwọn Àkọ́kọ́ Àwọn Àkọ́kọ́ Àwọn Àkọ́kọ́)

Fi àkọlé rẹ pamọ́ sí àwọn àmì-ìwé SSML fún ìdáràn:

<speak><prosody rate="slow">Slow speech</prosody></speak>

Àwọn Àmì-ìwé Ìjánu-ìró / Ìyárà

Àwọn Àmì-ìwé tí àwọn ìṣàmúlò-ètò tí a yàn gbọ́ - tẹ̀ láti fi ọkan sínú àkọ́lé rẹ̀ nínú àwọn ààyè-iṣẹ́ tí o bá jẹ́:

Àwọn Àwọn Àkọlé

Àwọn àwọn ìṣàfarawé àwọn àwọn ìṣàfarawé àwọn (ọrọ = ìṣàfàlì):

Ìjánu-ìsún 0

-12 +12

Àwọn ìṣàmúlò-ètò

Àwọn àmì-ìwé

Àwọn

Ìgúnrégé àwọn ààtòjọ-ẹ̀yàn

Ìjánu-ìṣàmúlò-ètò 1.0x

0.5x 2.0x

Free pẹlu Piper, VITS, MeloTTS

Àwọn àwòrán tí o ti ṣẹ̀dà tí o bá han níbẹ̀. Yan àwọn àwòrán, tẹ̀lẹ̀ àkọlé, ki o si tẹ̀ Ṣẹ̀dà.

Ààyè-iṣẹ́ Darwin TTS

Darwin-TTS-1.7B-Cross by FINAL-Bench is a research variant of Qwen3-TTS-1.7B with an unusual construction: 84 of its talker-FFN tensors (about 8.6% of them) are blended at a 3% ratio with the matching tensors from Qwen3-1.7B-Base, all without any retraining. The result is a model that produces noticeably crisper cross-lingual voice cloning across Korean, English, Japanese, and Chinese — its four core languages. It operates in zero-shot voice-clone mode, needing only about three seconds of reference audio to capture a speaker. Darwin is best suited to transferring a single reference voice across those four languages, for example dubbing or multilingual narration with consistent speaker identity.

Tí o dara jù fún: Cross-lingual voice cloning between English / Korean / Japanese / Chinese with a single reference voice

Wá Gbogbo àwòrán Darwin TTS Àwọn àwòrán

Nínú àwọn ìṣàfarawé

Àwọn Àkọlé: FINAL-Bench
Àwọn Ààyè-iṣẹ́: Apache 2.0
Àwọn àwọn ààyè-iṣẹ́: standard
Ìjánu-ìṣàmúlò-ètò: medium
Ìṣàfarawé àwọn àmì-ìwé: Yà
Àwọn: English, Korean, Japanese, Chinese
Àwọn àyọkà ìpele: 2000

Darwin TTS Àwọn àwòrán

Default

English

Àwọn ìpéwọ̀n Neutral

Default (Chinese)

Chinese

Àwọn ìpéwọ̀n Neutral

Default (Japanese)

Japanese

Àwọn ìpéwọ̀n Neutral

Default (Korean)

Korean

Àwọn ìpéwọ̀n Neutral

Darwin TTS Àwọn Àtòjọ-ẹ̀yàn

Darwin starts from Qwen3-TTS-1.7B but blends a small fraction of its talker-FFN weights with the matching weights from the Qwen3-1.7B base language model. This training-free blend sharpens cross-lingual voice cloning rather than changing the base voices.

English, Korean, Japanese, and Chinese. The FINAL-Bench release specifically markets its cross-lingual blend for those four, and the deployed model ships voices for them.

About three seconds. It works in zero-shot mode, so no fine-tuning or training is required — you provide a short reference clip and it generates new speech in that voice.

← Gbogbo àwọn ìrànwọ́

Darwin TTS TTS

O fẹ́ TTS.ai? Fì sọ̀kalẹ̀ fún àwọn ọrẹ̀ rẹ̀!

Ààyè-iṣẹ́ Darwin TTS

Nínú àwọn ìṣàfarawé

Darwin TTS Àwọn àwòrán

Default

Default (Chinese)

Default (Japanese)

Default (Korean)

Darwin TTS Àwọn Àtòjọ-ẹ̀yàn

How is Darwin TTS different from Qwen3-TTS?

Which languages does Darwin TTS handle?

How much reference audio does Darwin need to clone a voice?