Qwen Open-Sources Qwen3-TTS with Voice Cloning and Real-Time Streaming
Alibaba's Tongyi Qianwen team has open-sourced the Qwen3-TTS series, featuring 1.7B and 0.6B parameter models. Built on a novel 12Hz multi-codebook speech tokenizer and dual-track modeling, it achieves ultra-low latency streaming generation (first packet after one character) along with high-fidelity voice cloning, design, and control across 10 major languages. Code, weights, and demos are now available on GitHub and Hugging Face, with an API option for immediate testing.
Related Entities
Full Report
Contents17 sections
Github HuggingFace Huggingface Demo ModelScope Demo Paper
Qwen3-TTS 是由Qwen开发的一系列功能强大的语音生成,全面支持音色克隆、音色创造、超高质量拟人化语音生成,以及基于自然语言描述的语音控制,为开发者与用户提供最全面的语音生成功能。依托创新的 Qwen3-TTS-Tokenizer-12Hz 多码本语音编码器,Qwen3-TTS 实现了对语音信号的高效压缩与强表征能力,不仅完整保留副语言信息和声学环境特征,还能通过轻量级的非 DiT 架构实现高速、高保真的语音还原。Qwen3-TTS 采用 Dual-Track 双轨建模,达成了极致的双向流式生成速度,首包音频仅需等待一个字符。Qwen3-TTS 多码本全系列模型均已开源,包含1.7B和0.6B两种尺寸,1.7B可以达到极致性能,具有强大的控制能力,0.6B均衡性能与效率。模型覆盖 10 种主流语言(中文、英文、日语、韩语、德语、法语、俄语、葡萄牙语、西班牙语、意大利语)及多种方言音色,满足全球化应用需求。同时,模型具备强大的上下文理解能力,可根据指令和文本语义自适应调整语气、节奏与情感表达,并对输入文本噪声的鲁棒性有显著提升。目前已经在Github上开源同时也可通过Qwen API体验。
模型列表#
1.7B模型#
| 模型 | 功能 | 语种支持 | 流式生成 | 指令控制 |
|---|---|---|---|---|
| Qwen3-TTS-12Hz-1.7B-VoiceDesign | 根据用户输入描述进行音色创造。 | 中、英、日、韩、德、法、俄、葡萄牙、西班牙、意大利 | ✅ | ✅ |
| Qwen3-TTS-12Hz-1.7B-CustomVoice | 根据用户输入指令对目标音色进行风格控制;支持9个精品音色,涵盖多种性别、年龄、语种与方言组合。 | 中、英、日、韩、德、法、俄、葡萄牙、西班牙、意大利 | ✅ | ✅ |
| Qwen3-TTS-12Hz-1.7B-Base | 基座模型,可以根据用户输入音频进行3s快速音色克隆,可用于FT到其他模型。 | 中、英、日、韩、德、法、俄、葡萄牙、西班牙、意大利 | ✅ |
0.6B模型#
| 模型 | 功能 | 语种支持 | 流式生成 | 指令控制 |
|---|---|---|---|---|
| Qwen3-TTS-12Hz-0.6B-CustomVoice | 支持9个精品音色,涵盖多种性别、年龄、语种与方言组合。 | 中、英、日、韩、德、法、俄、葡萄牙、西班牙、意大利 | ✅ | |
| Qwen3-TTS-12Hz-0.6B-Base | 基座模型,可以根据用户输入音频进行3s快速音色克隆,可用于FT到其他模型。 | 中、英、日、韩、德、法、俄、葡萄牙、西班牙、意大利 | ✅ |
Qwen3-TTS 特性#
主要特点:
- 强大的语音表征:基于自研 Qwen3-TTS-Tokenizer-12Hz,实现语音信号的高效声学压缩与高维语义建模,完整保留副语言信息及声学环境特征,并可通过轻量级的非 DiT 架构实现高效、高保真语音还原。
- 通用的端到端架构:采用离散多码本 LM 架构,实现语音全信息端到端建模,彻底规避传统 LM+DiT 方案的信息瓶颈与级联误差,显著提升模型的通用性、生成效率与效果上限。
- 极致的低延迟流式生成:基于创新的 Dual-Track 混合流式生成架构,单模型同时兼容流式与非流式生成,最快可在输入单字后即刻输出音频首包,端到端合成延迟低至 97ms,满足实时交互场景的严苛需求。
- 智能的文本理解与语音控制:支持自然语言指令驱动的语音生成,灵活调控音色、情感、韵律等多维声学属性;同时深度融合文本语义理解,自适应调节语气、节奏、情感与韵律,实现“所想即所听”的拟人化表达。

模型性能#
我们对 Qwen3-TTS 在音色克隆、创造、控制等方面进行了全面评估,结果显示其在多项指标上都达到了SOTA性能。具体来说:
- 音色创造任务上,Qwen3-TTS-VoiceDesign 在 InstructTTS-Eval 中指令遵循能力和生成表现力都整体超越 MiniMax-Voice-Design 闭源模型,并大幅领先其余开源模型。
- 在音色控制任务上,Qwen3-TTS-Instruct 不仅具备单人多语言的泛化能力,平均词错率 2.34%;同时具备保持音色的风格控制能力,InstructTTS-Eval 取得了 75.4% 的分数;此外,也展现出卓越的长语音生成能力,一次性合成 10 分钟语音的中英词错率为 2.36/2.81%。
- 在音色克隆任务上,Qwen3-TTS-VoiceClone 在 Seed-tts-eval 上中英文克隆的语音稳定性表现上均超越 MiniMax 和 SeedTTS;在 TTS multilingual test set 上 10 个语项上取得了 1.835% 的平均词错误率和 0.789 的说话人相似度,超越 MiniMax 和 ElevenLabs;跨语种音色克隆也超越 CosyVoice3 位居 SOTA。


Tokenizer性能#
我们对 Qwen-TTS-Tokenizer 进行了语音重构评估,在LibriSpeech test-clean set的结果显示其在关键指标上都达到了的SOTA水平。具体来说,在感知语音质量评估(PESQ)中,Qwen-TTS-Tokenizer在宽带和窄带上分别取得了3.21和3.68的分数,大幅领先同类tokenizer。在短时客观可懂度(STOI)以及UTMOS上,Qwen-TTS-Tokenizer取得了0.96和4.16的分数,展现出卓越的还原质量。在说话人相似度上,Qwen-TTS-Tokenizer取了0.95的分数,显著超越对比模型,表明其近乎无损的说话人信息保留能力。

样例#
Qwen3-TTS-12Hz-1.7B-VoiceDesign#
音色创造#
Qwen3-TTS 支持通过自然语言描述生成定制化的音色形象。用户可以随意输入声学属性、人设描述、背景信息等自由描述,轻松创造出自己期望的声音形象。
| 控制类型 | 控制指令 | 合成文本 | 合成样例 |
|---|---|---|---|
| 声学属性控制 | 采用高亢的男性嗓音,语调随兴奋情绪不断上扬,以快速而充满活力的节奏传达信息。音量要足够响亮,近乎喊叫,以体现紧迫感。发音务必清晰精准、字字分明,让每个词都铿锵有力。整体表达需流畅自然、明亮生动,富有戏剧性,展现出外向、自信且张扬的个性,同时传递出一种威严而宏大的宣告语气,洋溢着满溢的激动之情。 | 好了各位,往后退,往后退!我有个天大的好消息要宣布:Qwen-TTS正式开源啦! | |
| gender: Male. pitch: Low male pitch with significant upward inflections for emphasis and excitement. speed: Fast-paced delivery with deliberate pauses for dramatic effect. volume: Loud and projecting, increasing notably during moments of praise and announcements. age: Young adult to middle-aged adult. clarity: Highly articulate and distinct pronunciation. fluency: Very fluent speech with no hesitations. accent: British English. texture: Bright and clear vocal texture. emotion: Enthusiastic and excited, especially when complimenting. tone: Upbeat, authoritative, and performative. personality: Confident, extroverted, and engaging. | Nine different, exciting ways of cooking sausage. Incredible. There were three outstanding deliveries in terms of the sausage being the hero. The first dish that we want to dissect, this individual smartly combined different proteins in their sausage. Great seasoning. The blend was absolutely spot on. Congratulations. Please step forward. Natasha. | ||
| 展现出悲苦沙哑的声音质感,语速偏慢,情绪浓烈且带有哭腔,以标准普通话缓慢诉说,情感强烈,语调哀怨高亢,音高起伏大。 | 皇上啊!臣妾一片真心可昭日月,为何您竟信那毒妇谗言,将我打入冷宫?这心……比雪还凉啊…… | ||
| gender: Male. pitch: Artificially high-pitched, slightly lowering after the initial laugh. speed: Rapid during the laugh, then slowing to a deliberate pace. volume: Loud laugh transitioning to a standard conversational level. age: Young adult to middle-aged, performing a character voice. clarity: Clear and distinct articulation. fluency: Fluent delivery without hesitation. accent: American English. texture: Slightly strained and somewhat nasal quality. emotion: Forced amusement shifting to feigned resignation. tone: Initially playful, then shifts to a slightly put-upon tone. personality: Theatrical and expressive. | Good one. Okay, fine, I'm just gonna leave this sock monkey here. Goodbye. | ||
| 年龄控制 | 体现撒娇稚嫩的萝莉女声,音调偏高且起伏明显,营造出黏人、做作又刻意卖萌的听觉效果。 | 哥哥,你回来啦,人家等了你好久好久了,要抱抱! | |
| Speak as a sarcastic, assertive teenage girl: crisp enunciation, controlled volume, with vocal emphasis that conveys disdain and authority. | Blah, blah, blah. We're all very fascinated, Whitey, but we'd like to get paid. | ||
| 性别: 男性. 音高: 男性低沉音域,音高稳定. 语速: 语速稍快,节奏紧凑. 音量: 音量洪亮,力度强劲. 年龄: 中老年. 清晰度: 发音清晰,字句有力. 流畅度: 表达流畅,一气呵成. 口音: 标准普通话. 音色质感: 嗓音浑厚,略带沙哑感. 情绪: 严肃告诫,指令明确. 语调: 命令式语调,强调果断. 性格: 权威果断,不容置喙. | 把你所有的表情都藏在面具里,保持你的中性状态,不用表情,只用身体的语言,要记住,要学会藏。 | ||
| gender: Male. pitch: Low male pitch, generally stable. speed: Deliberate pace, slowing slightly after the initial exclamation. volume: Starts loud, then transitions to a projected conversational volume. age: Middle-aged adult. clarity: High clarity with distinct pronunciation. fluency: Highly fluent. accent: American English. texture: Resonant and slightly gravelly. emotion: Initially commanding, shifting to narrative amusement. tone: Authoritative start, moving to an engaging, descriptive tone. personality: Confident and performative. | Older gentleman, 110, maybe 111 years old, sort of a surly Elvis thing happening with him. He smiles like this. Seen him around? | ||
| 渐变控制 | 性别: 男性 音高: 男性低沉音区,偶有拔高. 语速: 初始平稳,后段因激动逐渐加快. 音量: 初始音量正常,后段逐渐提高至喊叫. 年龄: 中年男性. 清晰度: 吐字清晰,发音准确. 流畅度: 言语连贯,表达自然. 口音: 标准普通话发音. 音色质感: 音质略带粗砺,富有力量感. 情绪: 初始不耐烦,迅速转为恼怒斥责. 语调: 质问命令式,语带不悦与威慑. 性格: 急躁易怒,态度强硬. | 你在干什么?有什么好看的?喂!我叫你走,你在干什么?给我走啊! | |
| gender: Female. pitch: Mid-range female pitch, rising sharply with frustration. speed: Starts measured, then accelerates rapidly during emotional outburst. volume: Begins conversational, escalates quickly to loud and forceful. age: Young adult to middle-aged. clarity: High clarity and distinct articulation throughout. fluency: Highly fluent with no significant pauses or fillers. accent: General American English. texture: Bright and clear vocal quality. emotion: Shifts abruptly from neutral acceptance to intense resentment and anger. tone: Initially accepting, becomes sharply accusatory and confrontational. personality: Assertive and emotionally expressive when provoked. | Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! And thanks to you- | ||
| 拟人感 | 自然感的女声,语调活泼带笑意,模仿别人‘嘘’你时压低嗓音,就是平时聊天的感觉 | 我跟我闺蜜看电影就特别有画面感。就你知道吗,一紧张我就忍不住吃爆米花,吃得特别快,然后手里那杯可乐也跟着晃,差点就洒了,真的差一点点。然后旁边那个人就突然来一句,嘘——声音压得特别低。哎我当下那个情绪,既想笑又有点气,太尴尬了。 | |
| A relaxed, naturally expressive male voice in his late twenties to early thirties, with a moderately low pitch, casual speaking rate, and conversational volume; deliver lines with a light, self-deprecating tone, breaking into genuine, easygoing laughter at moments of embarrassment, while maintaining clear articulation and an overall warm, approachable clarity. | Yeah, so—uh—I’m a digital nomad, right? So… pretty much all my communication is just, like, texts and messages. And now, you know, there’s these AI agents that can, uh… reply for you? Which is—heh—convenient, sure, I guess? But also… kinda delicate, you know? Like, you’ll type something super short—like, “Yep, sounds good”—and it’ll turn that into this whole… warm, polished paragraph. Like, way nicer than I’d ever write myself. huh… ha Seriously, I sound like a Hallmark card all of a sudden. But then… once you outsource that… what’s the other person actually hearing? Are they hearing me… or just some… generic, friendly-bot voice? Man, that’s weird to even say out loud. | ||
| 背景信息 | 角色姓名:林怀岳 音色信息:音量洪亮,音域低沉,力度感强的中年男性声音。 身份背景:某国家重点科研项目首席顾问,年近七十的资深战略科学家。曾参与国家重大科技攻关工程,历经数十年风雨,见证了从落后追赶到自主创新的艰难历程。现任国家科技咨询委员会终身荣誉委员,仍坚持在一线培养青年人才,为国家战略发展建言献策。 外貌特征:身形挺拔,两鬓斑白,眉宇间刻着岁月沉淀的坚毅。常着深色中山装或简洁正装,眼神沉静而锐利,举手投足间自带威严与从容。 性格特质:意志如钢,信念坚定,面对挑战从不退缩;胸怀家国,心系民族未来,将个人命运与国家兴衰紧密相连;严谨自律,言出必行,话语中充满责任感与历史担当;外冷内热,表面严肃,实则对后辈寄予厚望,甘为人梯。 人生信条:“我们这一代人,不是为了站在光里,而是为了把路铺到光里。” | 有些事,只要国家需要,就得有人扛起来。 我们那一代人,是背着泥土铺路的; 你们要做的,是让这条路,通向星辰大海。 | |
| Character Name: Marcus Cole Voice Profile: A bright, agile male voice with a natural upward lift, delivering lines at a brisk, energetic pace. Pitch leans high with spark, volume projects clearly—near-shouting at peaks—to convey urgency and excitement. Speech flows seamlessly, fluently, each word sharply defined, riding a current of dynamic rhythm. Background: Longtime broadcast booth announcer for national television, specializing in live interstitials and public engagement spots. His voice bridges segments, rallies action, and keeps momentum alive—from voter drives to entertainment news. Presence: Late 50s, neatly groomed, dressed in a crisp shirt under studio lights. Moves with practiced ease, eyes locked on the script, energy coiled and ready. Personality: Energetic, precise, inherently engaging. He doesn’t just read—he propels. Behind the speed is intent: to inform fast, to move people to act. Whether it’s “text VOTE to 5703” or a star-studded tease, he makes it feel immediate, vital. | Lot being you watching. 1-866-IDLE-03 for JPL. That's 1-866-436-5703. Or text the word VOTE to 5703. Diana DeGarmo's next with more from the movies right after this brief intermission on American Idol. |
音色复用#
用户也可将 Qwen3-TTS 创建的音色进行持久存储和重复调用,生成生动自然的多轮次多角色长篇章对话。
| 控制指令 | 合成文本 | 合成样例 |
|---|---|---|
| "旁白": "声音特征沉稳、客观、略带叙事感的女播音腔,普通话标准,语速适中,带有轻微的环境氛围渲染,语调平缓但富有感染力,在关键情节时稍作停顿,增强画面感。情感冷静旁观,偶尔带一丝微妙的反讽" "小林": "25岁男性上班族,声音清亮但时常犹豫,语速时快时慢,紧张时会轻微结巴。情绪波动明显,从低声呢喃到突然激动再到自我怀疑的叹气。肢体语言丰富,经常无意识的小动作" "御姐": "模拟成熟性感的御姐音色,声音略带磁性且沉稳,语速不快不慢,语调充满自信和一丝挑逗,尾音可以稍微拖长并上扬,给人一种游刃有余的掌控感。" | 旁白: 小林今天第三次走神了。酒吧昏黄的灯光晃得他心跳加速,而吧台对面那个红唇微扬的女人,正用指尖轻轻摩挲着酒杯边缘。 御姐: 小弟弟,有兴趣陪姐姐喝一杯吗? 小林: 啊?我、我……我其实不太会喝酒…… 旁白: 他的手指无意识地抠着杯沿,喉结上下滚动,像被什么无形的东西掐住了呼吸。 御姐: 不会喝?那正好——姐姐教你。这杯莫吉托,甜得刚好,就像你刚才偷看我的眼神。 小林: 我、我没偷看!……好吧,看了一眼。就一眼! 旁白: 他猛地坐直,又立刻缩回肩膀,仿佛那句话烫伤了自己的嘴。 御姐: 紧张什么?你连坐姿都在发抖……要不要靠过来一点?这里太吵了。 小林: 靠过去?可、可我们才第一次见面……你都不认识我…… 御姐: 名字不重要,感觉才重要。......而我感觉……你有点可爱。 旁白: 小林的耳朵瞬间红透,连耳后那颗小痣都像在发烫。他想逃,脚却像钉在了高脚凳上。 小林: 可爱?没人这么说过我……他们都说我太闷,连朋友圈都发不出手…… 御姐: 那现在呢?敢不敢发一条——'今晚,和一个危险又迷人的姐姐喝了一杯'? 小林: ……我连配图都不敢选。你笑起来太……太有杀伤力了。 御姐: 那就别发了。有些故事,只适合藏在两个人的记忆里——比如,接下来你打算请我跳支舞吗? 旁白: 他张了张嘴,没发出声音。但这一次,他没有低头,而是轻轻推开了那杯没动过的苏打水,朝她伸出了手。 | |
| ""Lucas": "Male, 17 years old, tenor range, gaining confidence - deeper breath support now, though vowels still tighten when nervous" "Mia": "Female, 16 years old, mezzo-soprano range, softening - lowering register to intimate speaking voice, consonants softening" | Lucas! You dropped your... uh... calculus notebook? I mean, I think it's yours? Maybe? Mia wow, my mortal enemy - Mr. Thompson's problem sets. Thanks for rescuing me from that F. Lucas problem! I actually... kinda finished those already? If you want to compare answers or something... Mia this your sneaky way of saying you want to study together, Lucas? Because I saw you staring during lab partners sign-up. Lucas? No! I mean yes but not like... I just think you're... your titration technique is really precise! Mia's the nerdiest compliment I've ever gotten. Tell you what - help me survive pre-calc and I'll teach you how to actually flirt. Lucas, harsh. And here I thought my titration line was smooth. Mia was adorable. Like when you tripped over your shoelaces in the hall yesterday. Or that time you— Lucas okay! I get it, I'm a disaster. So... library after school? I'll bring the graphing calculators? Mia if you promise not to spill coffee on my notes again... though I guess watching you panic-clean was pretty cute. |
Qwen3-TTS-12Hz-1.7B-CustomVoice#
音色控制#
Qwen3-TTS 进行特定说话人微调后,可以在保持目标音色的同时,继承基模型的风格控制能力以及单人多语能力。
| 控制类型 | 音色 | 控制指令 | 合成文本 | 合成样例 |
|---|---|---|---|---|
| 单属性控制 | 甜茶 Ryan | spoke with a very sad and tearful voice. | She said she would be here by noon. | |
| Very happy. | ||||
| 用特别愤怒的语气说 | ||||
| 请特别小声的悄悄说 | ||||
| Speaking at an extremely slow pace | ||||
| 音调低沉 | ||||
| 多属性控制 | 十三 Vivan | 性别: 女性声音. 音高: 女性中高音区,语调富于变化. 语速: 语速明快,偶有加速. 音量: 正常交谈音量,笑声响亮. 清晰度: 吐字清晰,发音标准. 流畅度: 表达流畅自如. 口音: 普通话. 音色质感: 音色明亮,略带爽朗. 情绪: 愉悦友好,伴随爽朗笑意. 语调: 语调上扬活泼,疑问时尤为明显. 性格: 外向开朗,热情健谈. | 就算你自己不想治,你也得考虑考虑别人的感受吧。我们这些朋友的感受你不在乎无所谓,那你家人呢?你家人的感受你难道一点都不在乎吗! | |
| 以极度悲伤、带着明显哭腔的语气,用较小的音量缓缓诉说,语速缓慢,仿佛每一个字都承载着沉重的痛楚,声音颤抖而压抑,吐字虽轻却清晰可辨,透出深藏心底的哀伤与无助。 | ||||
| 保持青年女性的声线特征,展现出一种清亮且略具紧迫感的音色,语速从平稳开始在叙述过程中逐渐加快,音量在情绪波动时增加,语调在句末调高以强调劝告的语气。 | ||||
| 单人多语泛化 | 十三 Vivan | 在语速偏快的情况下流畅自然地表达,音质清亮,音调略高,吐字清晰标准,给人一种开心愉悦的感觉。 | (韩语)안녕하세요, 오늘은 어떤 용건입니까? | |
| A deep, rich, and solid vocal register characteristic of a middle-aged woman, with full and powerful volume. Speech is delivered at a steady pace, articulation clear and precise, with fluent and confident intonation that rises slightly at the end of sentences. | (日语)こんにちは、本日はどのようなご用件でしょうか? | |||
| 语音应表现为直率且略显主观强势的中年女性,音色略带尖锐感,流畅表达中偶尔断句以凸显语气,情绪略带不满,音量随情感激动略有增强。 | (中文方言-四川话)我早就该下班了,就是跟你说我这事情干不完,我现在走不脱。 |
音色列表#
本次Qwen3-TTS共开源了9个音色,涵盖多种性别、年龄、语种与方言组合,满足不同场景下的个性化语音生成需求。
| 音色 | 语种 | 合成文本 | 合成样例 |
|---|---|---|---|
| 苏瑶 Serena | 中文 | 其实我真的有发现,我是一个特别善于观察别人情绪的人。 | |
| 福伯 Uncle Fu | 中文 | 其实我真的有发现,我是一个特别善于观察别人情绪的人。 | |
| 十三 Vivian | 中文 | 其实我真的有发现,我是一个特别善于观察别人情绪的人。 | |
| 艾登 Aiden | 英文 | Then by the end of the movie, when Dorothy clicks her heels and says, “There’s no place like home,” I got a little bit teary, I’ll admit. You know, I don’t even know why—I just, I just felt. | |
| 甜茶 Ryan | 英文 | Then by the end of the movie, when Dorothy clicks her heels and says, “There’s no place like home,” I got a little bit teary, I’ll admit. You know, I don’t even know why—I just, I just felt. | |
| 小野杏 Ono Anna | 日语 | やばい、明日のプレゼン資料まだ完成してない… 助けて! | |
| 素熙 Sohee | 韩语 | 야, 오늘 점심에 뭐 먹을지 생각해 봤어? 근처에 새로 생긴 분식집 어때? | |
| 晓东 Dylan | 中文方言-北京话 | 我们就在山上啊,就是其实也没什么,就是在土坡上跑来跑去,然后谁捡个那个嗯比较威风的棍儿,完了我们就就瞎打,呃要不就是什么掏个洞啊什么的。 | |
| 程川 Eric | 中文方言-四川话 | 你龟儿太过分了,把我的东西都搞坏了,还晓不晓得认错,硬是要把我整冒火你才安逸嗦,莫再烦老子爬球开。 |
Qwen3-TTS-12Hz-1.7B-Base#
音色克隆#
| 控制类型 | 参考音频 | 合成文本 | 合成样例 |
|---|---|---|---|
| 中文克隆 | 你眼中的太阳,只是我指间的玩物。 | ||
| 祝您在马年里事业一马当先,业绩万马奔腾,在新的一年里快马加鞭,再创辉煌! | |||
| 英文克隆 | In the absence of confiscation, the Portuguese inquisitors were not earnest in tracing the heresies of ancestors or in following up the records of fugitives. | ||
| An ideal harmonious society. Humanism, is a lighthouse on this way to guide us in case we are getting lost. | |||
| 跨语种克隆 | (日语)要約すれば、全米国民のためにアメリカを再興するという使命を、我々は開始したのである。 | ||
| (韩语)광활한 우주 속에, 지구라고 불리는 아름다운 푸른 행성이 있습니다. | |||
| 文本鲁棒性 | Qwen-TTS 是支持音色克隆、生成、控制的开源语音合成模型,不仅支持多语言multilingual,还支持各种复杂文本,如pin1 yin1,特殊符号等(◍•͈⌔•͈◍);能读出各种生僻字詞。快来试试吧! | ||
| I am solving the equation: x = [-b ± √(b²-4ac)] / 2a? Nobody can — it's a disaster (◍•͈⌔•͈◍), very sad! |
Qwen-TTS-Tokenizer-12Hz#
音频重构#
| 重构类型 | 原始样例 | 重构样例 |
|---|---|---|
| 方言重构 | ||
| 歌声重构 | ||
| 副语言重构 | ||
| 背景声重构 |