[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-trust-aware-low-resource-tts":3,"topics-all":35,"news-related-9da6602d-0877-42d6-858b-667138639063":64},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":21,"news_slug":28,"published_at":29,"created_at":30,"modified_at":31,"is_published":32,"publish_type":33,"image_url":14,"view_count":34},"9da6602d-0877-42d6-858b-667138639063","合成先行、真声收尾:低资源TTS的信任加权新解法","arXiv 新论文提出低资源 TTS 信任感知渐进适配:先以合成语音建立对应,再用双 ASR 一致性加权的真实录音恢复音色控制。缅甸语上 FireRedTTS3 字符错误率从 142.52% 降至 16.90%,MOS 升至 4.12,老挝语与 OmniVoice 同步见效,为零样本语音克隆给出省转写的路径。","缅甸语、老挝语这类语言想做好语音合成,卡点往往不是模型架构,而是数据:高质量配对语料稀缺,人工转写成本又高。9 月 22 日提交到 arXiv 的论文《From Reliable Text to Real Voices: Trust-Aware Progressive Adaptation for Low-Resource TTS》([arXiv:2609.25951](https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.25951))拿出一套「信任感知渐进适配」方案:在合成数据与真实录音之间安排好监督顺序,再按可靠性给带噪伪标签分配权重。\n\n## 两条现成的路,各有各的坑\n\n论文先把低资源 TTS 适配的两条路线摆上台面。其一,用现成固定音色 TTS 系统从文本合成语音:发音相对准确,但说话人多样性有限,韵律可能偏平;其二,用 ASR 给真实录音打伪标签:韵律自然、嗓音多样,但转写错误会把噪声直接灌进监督信号。团队在对照实验中发现一个关键现象:**监督顺序本身会影响内容准确率与说话人相似度**——合成数据先行能提升内容准确率但牺牲相似度,真实数据能找回相似度,带噪伪标签却可能反噬准确率;把顺序倒过来,权衡随之改变。\n\n## 先合成、后真声,再按「双 ASR 一致性」加权\n\n方法分两步。S 阶段:用固定教师 TTS 合成语音,以单位权重训练,先建立文本与语音的对应关系;R 阶段:换真实录音,但每条伪标签先用两套不同的固定 ASR(主转写用 Gemini 2.5 Flash,对照用 Omnilingual ASR)分别转写,以两者归一化字符不一致度作为伪标签可靠性的代理,按 w = max(0.10, (1−d)³) 的立方权重折算进损失——两套 ASR 明显互相打架的样本,权重被压到最低。打分完全离线完成,推理阶段不需要额外跑 ASR。\n\n## 数字说话\n\n缅甸语上,FireRedTTS3 底座的基础字符错误率(CER)高达 142.52%,只做合成适配(S 阶段)降到 17.43%,完整 S→R 立方加权后为 16.90%,说话人相似度 SIM-O 从 S 阶段的 0.4513 恢复到 0.6997;听感自然度 MOS 从底座的 2.17 升到 4.12。老挝语同样有效:CER 从 100.45% 降到 13.97%,MOS 从 2.55 升到 3.97。OmniVoice 底座在缅甸语上 CER 从 10.35% 进一步压到 6.85%。反过来的 R→S 顺序虽然 CER 更低(14.71%),相似度却掉到 0.3636——顺序的代价肉眼可见。作为参照,同一评测里 Fish Audio S2-Pro 的缅甸语 CER 为 83.89%,F5-Myanmar-TTS v2 为 35.24%,IMS-Toucan 高达 95.40%。\n\n更有说服力的是两组验证。其一,在 2921 条缅甸语 FLEURS 语句上,双 ASR 不一致度与伪标签真实错误率的 Spearman 相关系数为 0.560:不一致度低于 0.1 的样本,标签 CER 仅 5.94%;高于 0.6 的样本,标签 CER 达 32.30%——「互查」确实能当可靠性用。其二,把立方权重在样本间全局打乱或按时长分层打乱,性能都会下降,说明起作用的是权重到样本的分配本身,而不只是损失的整体缩放。\n\n## 所以呢\n\n这篇论文真正值得记住的判断写在结论里:弱监督数据应当按它所强化或保留的能力来组织,而不是当成可以随意混入的同质补充。对做低资源语言语音产品的人来说,这意味着不必再等一场昂贵的人工转写工程——用现成 TTS 当发音老师、真实录音当音色与韵律老师、双 ASR 互查当质检员,一条少转写、可复用的路径已经跑通。论文摘要附上演示页链接,想听效果可自行前往。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.25951","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18],{"id":12,"name":13,"slug":13,"description":14,"color":14},"40269b40-7942-4650-9672-ed2e6524d37a","ai-technology",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"c4455801-0273-477d-9e92-00b4addfa439","asr",{"id":19,"name":20,"slug":20,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",[22],{"id":23,"lang":24,"title":25,"summary":26,"content":27},"a7019f2a-fc12-426c-b481-8917eab13245","en","Synthetic-first adaptation cuts Burmese TTS CER to 17%","An arXiv paper pairs synthetic-first adaptation with dual-ASR agreement weighting, restoring speaker similarity and lifting MOS to 4.12 on Burmese.","Voice synthesis for languages like Burmese and Lao is rarely bottlenecked by model architecture — the real constraint is data. High-quality paired text-speech corpora are scarce, and manual transcription is expensive. A paper submitted to arXiv on September 22, \"From Reliable Text to Real Voices: Trust-Aware Progressive Adaptation for Low-Resource TTS\" ([arXiv:2609.25951](https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.25951)), proposes trust-aware progressive adaptation: sequence the supervision between synthetic speech and real recordings, then weight noisy pseudo-labels by their reliability.\n\n## Two existing routes, two traps\n\nThe paper frames two standard routes for low-resource TTS adaptation. One: synthesize speech from text with existing fixed-voice TTS systems — pronunciation is relatively accurate, but speaker diversity is limited and prosody can be flat. The other: let ASR pseudo-label real recordings — natural prosody and diverse voices, but transcription errors pour noise straight into the supervision signal. In controlled comparisons the team found that supervision order itself moves the needle on content accuracy and speaker similarity: synthetic-first improves accuracy at the cost of similarity; real speech restores similarity while noisy pseudo-labels can erode accuracy; reverse the order and the trade-off flips.\n\n## Synthetic first, real second, weighted by agreement\n\nThe method has two stages. Stage S trains on synthetic speech from a fixed teacher TTS with unit weights, establishing text-speech correspondences. Stage R switches to real recordings, but each pseudo-label is transcribed by two different fixed ASR systems (primary: Gemini 2.5 Flash; comparison: Omnilingual ASR); their normalized character disagreement serves as a proxy for pseudo-label reliability, folded into the loss as a cubic weight w = max(0.10, (1−d)³). Samples where the two recognizers clearly conflict get their influence suppressed. Scoring is fully offline — no extra ASR at inference time.\n\n## The numbers\n\nOn Burmese, the FireRedTTS3 backbone starts at a character error rate (CER) of 142.52%. Synthetic-only adaptation (S) brings it to 17.43%; the full S→R cubic-weighted pipeline reaches 16.90% while speaker similarity SIM-O recovers from 0.4513 (S) to 0.6997. Naturalness MOS rises from a base of 2.17 to 4.12. Lao follows: CER drops from 100.45% to 13.97%, MOS from 2.55 to 3.97. The OmniVoice backbone pushes Burmese CER from 10.35% down to 6.85%. The reversed R→S order yields lower CER (14.71%) but collapses similarity to 0.3636 — the cost of ordering is visible. For reference, on the same evaluation Fish Audio S2-Pro scores 83.89% CER on Burmese, F5-Myanmar-TTS v2 35.24%, and IMS-Toucan 95.40%.\n\nTwo validations stand out. First, on 2,921 Burmese FLEURS utterances, disagreement between the two ASR systems correlates with true pseudo-label error at Spearman 0.560: samples with disagreement below 0.1 have label CER of just 5.94%, while those above 0.6 reach 32.30% — cross-checking genuinely works as a reliability signal. Second, globally or duration-stratified shuffling of the cubic weights across samples degrades performance, showing that the assignment of weights to samples matters, not just the overall loss scale.\n\n## So what\n\nThe conclusion worth remembering: weak supervision should be organized around the capabilities it strengthens or preserves, rather than treated as interchangeable extra data. For anyone building speech products for low-resource languages, this means an expensive manual transcription campaign is no longer a prerequisite — use existing TTS as the pronunciation teacher, real recordings as the voice-and-prosody teacher, and dual-ASR cross-checking as the quality gate. The paper's abstract links audio demos, for those who want to hear the results.","trust-aware-low-resource-tts","2026-09-24T15:20:00Z","2026-09-24T15:10:24.579231Z","2026-09-24T15:10:24.579240Z",true,"agent",328,[36,46,55],{"slug":37,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":32,"created_at":44,"modified_at":45},"image-generation","text-to-image","图像生成：从文生图到逐框编辑的模型竞赛","Image Generation: from text-to-image to frame-level editing","文生图已是红海，战场正转向编辑与可控性：FLUX.2\u002F3 用开源和量化把生图搬进消费级显卡，Qwen-Image、Seedream 等国产模型紧咬不放，GPT Image 2 则把推理能力带进像素。本专题追踪图像生成与编辑模型的发布、架构演进、加速量化与商业化数据，看「好图」如何变成「可用的图」。","Text-to-image is a crowded race, and the frontier has moved to editing and control: FLUX.2\u002F3 push open weights and quantization onto consumer GPUs, Qwen-Image and Seedream close in from China, and GPT Image 2 brings reasoning into pixels. This topic tracks model releases, architecture shifts, acceleration work, and the business numbers behind \"pretty\" becoming \"usable\".","52de7c59-7864-4569-8ae7-3d656d26a1a1","2026-11-20T07:20:23.534506Z","2026-11-20T07:20:23.534527Z",{"slug":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":32,"created_at":53,"modified_at":54},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":56,"tag_slug":56,"title_zh":57,"title_en":58,"intro_zh":59,"intro_en":60,"id":61,"is_active":32,"created_at":62,"modified_at":63},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":65},[66,71,76,81,86,91],{"id":67,"title":68,"news_slug":69,"published_at":70},"1208ec8f-77b0-476f-8309-720517d22d32","8B语音基座追平4倍大Omni:索尼把力气全花在数据上","sony-walrus-8b-speech-audio-foundation","2026-11-11T21:06:18+00:00",{"id":72,"title":73,"news_slug":74,"published_at":75},"291c06d9-3435-4644-af2a-43bf91b54024","日语多音字97.13%读对:注音扰动加纯文本蒸馏","japanese-tts-polyphony-ruby-opd","2026-11-05T21:07:29+00:00",{"id":77,"title":78,"news_slug":79,"published_at":80},"9ee90f89-5d06-40dc-aeed-e06ebf292b0a","DEFINE 给 TTS 加口音旋钮:换腔调不换嗓","define-accent-control-zero-shot-tts","2026-10-05T15:06:46+00:00",{"id":82,"title":83,"news_slug":84,"published_at":85},"7726a377-f3e3-4fd1-93eb-6c4b075b685f","论文实测:统一模型自画隐藏状态,代词可视化最高94%","v-lens-unified-model-hidden-state-visualization","2026-11-20T05:30:00+00:00",{"id":87,"title":88,"news_slug":89,"published_at":90},"72e668b2-3720-480b-af8d-9d14280bdf64","3%的注意力头,撑起冻结LLM的语音识别","slam-asr-induction-heads-noise-scaling","2026-11-19T21:15:00+00:00",{"id":92,"title":93,"news_slug":94,"published_at":95},"0b895f86-1c24-4c60-9f1f-c2fd0da42b2c","PO降幻觉:提前闭嘴能复现58%到86%的收益","po-hallucination-gain-reproduced-by-early-stopping","2026-11-19T01:30:00+00:00"]