{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-28T06:20:51.496Z","headline":"Inflect-Micro-v2 发布：仅 936 万参数实现完整语音模型","description":"Inflect-Micro-v2 以仅 936 万个参数实现了完整的语音模型功能。该模型在 HuggingFace 上发布，参数规模远小于主流语音模型，展示了极低参数量下仍能保持语音生成能力的可行性。","url":"https://www.aioga.com/news/cms1i68ky007uroxo9fq348kt/","mainEntityOfPage":"https://www.aioga.com/news/cms1i68ky007uroxo9fq348kt/","datePublished":"2026-07-26T07:38:53.861Z","dateModified":"2026-07-26T07:38:53.861Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://huggingface.co/owensong/Inflect-Micro-v2","https://aihot.virxact.com/items/cms1i68ky007uroxo9fq348kt"],"canonicalUrl":"https://www.aioga.com/news/cms1i68ky007uroxo9fq348kt/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Inflect-Micro-v2 以仅 936 万个参数实现了完整的语音模型功能。 Aioga 将其归入「模型更新」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/cms1i68ky007uroxo9fq348kt/","dateCreated":"2026-07-26T07:38:53.861Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"huggingface.co source article","url":"https://huggingface.co/owensong/Inflect-Micro-v2","datePublished":"2026-07-26T07:38:53.861Z","provider":{"@type":"Organization","name":"huggingface.co","url":"https://huggingface.co/owensong/Inflect-Micro-v2"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cms1i68ky007uroxo9fq348kt","datePublished":"2026-07-26T07:38:53.861Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cms1i68ky007uroxo9fq348kt"}}],"aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","originalPublisher":{"name":"huggingface.co","url":"https://huggingface.co/owensong/Inflect-Micro-v2"},"article":{"id":"cms1i68ky007uroxo9fq348kt","slug":"cms1i68ky007uroxo9fq348kt","url":"https://www.aioga.com/news/cms1i68ky007uroxo9fq348kt/","title":"Inflect-Micro-v2 发布：仅 936 万参数实现完整语音模型","title_en":"Inflect-Micro-v2：仅需936万个参数即可实现完整的语音模型","summary":"Inflect-Micro-v2 以仅 936 万个参数实现了完整的语音模型功能。该模型在 HuggingFace 上发布，参数规模远小于主流语音模型，展示了极低参数量下仍能保持语音生成能力的可行性。","source":"Hacker News 热门（buzzing.cc 中文翻译）","sourceUrl":"https://huggingface.co/owensong/Inflect-Micro-v2","aiHotUrl":"https://aihot.virxact.com/items/cms1i68ky007uroxo9fq348kt","publishedAt":"2026-07-26T07:38:53.861Z","category":"模型更新","score":46,"selected":false,"articleBody":["：/owensong/Inflect-Micro-v2/blob/main/assets/inflect-v2-repository-hero.png","Complete local text-to-waveform speech synthesis under 10M parameters. Fixed-voice English TTS with deterministic seeds, long-text handling, and CPU or CUDA inference.","I built and funded Inflect v2 independently. If this release finds a real audience, I would like to continue the project with a broader v3, which might include things like more langauges, voices, and stability improvements. If the model is useful to you, leaving a like on Hugging Face genuinely helps more people discover it.","：https://huggingface.co/spaces/owensong/Inflect-v2 ：https://github.com/owenawsong/Inflect ：https://huggingface.co/owensong/Inflect-Nano-v2 ：https://discord.gg/CVJYedvzvp ：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/EVALUATION.md","9,356,513 deployable parameters · 37.53 MB FP32 · 24 kHz mono output","Inflect v2 uses one public API across two sizes: Micro prioritizes quality below 10M parameters; Nano prioritizes footprint below 4M.","These are held-out text generations, not reconstructions of training audio. Each transcript is shown exactly as passed to the public frontend.","No single metric captures TTS quality. Inflect v2 reports human preference , predicted naturalness , multi-ASR intelligibility , complete footprint , and runtime separately rather than compressing them into one unverifiable score.","The headline row always refers to Inflect-Micro-v2 . Detailed competitor results and protocol boundaries are kept visible below.","Comparison set. Results include KittenTTS Nano：https://huggingface.co/KittenML/kitten-tts-nano-0.8, Piper Low：https://huggingface.co/rhasspy/piper-voices, and Supertonic 3：https://huggingface.co/Supertone/supertonic-3, established compact or local TTS baselines with larger deployable weight footprints than both Inflect releases. Weight sizes are compared at package level, and no single metric is treated as proof of overall superiority.","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/human-preference.svg","Inflect-Micro-v2 recorded a 66.2% preference rate (21 wins · 10 losses · 3 ties) in the final anonymous community study. Systems were hidden, left/right order was randomized, and ties count as half a win. This is descriptive community evidence, not formal MOS.","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/quality-vs-footprint.svg","The UTMOS22 run used 500 identical unseen prompts per voice. KittenTTS and Piper are equal-weight two-voice means; their observed voice ranges appear as whiskers. Supertonic 3-step is reported below the plotted range rather than flattening every other system.","Inflect-Micro-v2: 4.395 UTMOS22 , 95% bootstrap CI 4.381–4.408 . UTMOS22 is a learned predictor, not human MOS.","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/asr-consensus.svg","The headline score is the equal-weight mean of Qwen3-ASR and Nemotron 3.5 corpus WER for every system. Whisper is excluded consistently from the headline because it produced insertion-heavy hallucinations on a subset of otherwise intelligible Supertonic 8-step clips. It is not deleted: the complete three-ASR evidence remains below.","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/modern400-three-asr.svg","For Inflect-Micro-v2, the individual results are 2.52% Qwen3-ASR , 5.45% Nemotron 3.5 , and 2.73% Whisper large-v3 . The former three-model mean, 3.57% , is retained only as a descriptive audit value and is not used as the headline score.","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/asr-robustness.svg","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/category-semantic-wer.svg","These views are diagnostics, not additional leaderboards. They show where the recognizers disagree and which prompt categories still produce recoverable transcription errors.","Both Inflect releases synthesize comfortably faster than real time on CPU. The managed reference run used a Hugging Face CPU Upgrade instance (8 vCPU, 32 GB RAM) with four framework threads , end-to-end text-to-waveform timing, and 100 fixed Modern400 prompts. Three complete passes were recorded; the first cache-building pass was excluded and the table pools passes two and three.","These are package-level results from the public PyTorch runtime, not a claim that Inflect is the fastest compact TTS system. Hardware, frontend behavior, framework, compilation, and thread policy all affect small-model measurements.","The same managed CPU and four-thread policy were used for a shorter comparator pass: the identical 50-prompt prefix, repeated twice. KittenTTS and Piper are equal-work pooled across their two tested voices.","Because Inflect uses the larger 100-prompt steady-state run while comparator rows use the shorter 50-prompt confirmation pass, this table is deployment context rather than a perfectly matched speed leaderboard. Several comparators also use optimized ONNX runtimes, while the published Inflect benchmark above uses the canonical PyTorch runtime. The separately released Inflect ONNX path has not been substituted into those benchmark numbers.","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/model-footprint.svg","Voice variants sharing the same weights are merged. Inflect totals include the integrated waveform decoder.","Inflect-Micro-v2 is the quality-focused member of the family. Both models use the same public API and complete text-to-waveform packaging.","This uses the Hub's version-aware downloader and retrieves the complete repository. A Git clone also works, but hf download is the recommended path for ordinary model installation.","The result is a 24 kHz mono float32 waveform. Long input is split at punctuation-aware boundaries, synthesized chunk by chunk, and joined with controlled pauses.","The official verified FP32 export is published separately as Inflect-Micro-v2-ONNX ：https://huggingface.co/owensong/Inflect-Micro-v2-ONNX. It supports dynamic lengths, CPU/CUDA/DirectML provider selection, deterministic seeds, and the same long-text wrapper without importing PyTorch:","The neural model is split into duration.onnx and decode.onnx ; together they contain the complete learned text-to-waveform path. The English eSpeak-ng frontend remains CPU-side code. See the ONNX repository ：https://huggingface.co/owensong/Inflect-Micro-v2-ONNX for graph contracts, provenance, parity measurements, browser deployment, and re-export instructions.","Inflect v2 is a parameter-efficient VITS-family end-to-end text-to-waveform generator with an English phoneme frontend, monotonic alignment, stochastic latent synthesis, residual coupling flow, and an integrated alias-reduced neural waveform decoder.","The release describes the deployable architecture. Private corpus-construction and optimization details are not part of this open-weight package.","Long passages are punctuation-aware chunks, not one unlimited autoregressive pass. Chunk boundaries receive short pauses and edge fades. See docs/API.md ：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/API.md for waveform contracts and concurrency notes.","The release contains one fixed synthetic English voice. The package does not redistribute a real-speaker recording corpus, does not claim the voice as the identity of a real person, and requires no reference audio or external model at inference.","This release is inference-first. New-voice and new-language adaptation are not currently validated or supported . A new voice would replace the fixed speaker rather than add a selectable speaker; language adaptation also requires rebuilding normalization, phonemes, symbols, embeddings, and training data. See docs/DATA_AND_VOICE.md ：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/DATA_AND_VOICE.md and docs/FINETUNING.md ：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/FINETUNING.md.","Do not use the included voice to impersonate a real person, deceive listeners, or create fraudulent content. Disclose synthetic speech where the context could otherwise mislead. Users are responsible for applicable laws and the Apache-2.0 license.","Original Inflect code and weights are released under Apache-2.0. Bundled third-party components retain their own notices in THIRD_PARTY_NOTICES.md ：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/THIRD_PARTY_NOTICES.md. release_manifest.json records packaged file sizes and SHA-256 hashes.","Inflect v2 is an open-weight release. Deployable weights, inference code, frontend code, evaluation prompts, and release reports are public. The training corpus-generation pipeline, private filtering infrastructure, and full optimization recipe are not part of the public package.","Owen Song may share additional technical context privately for credible research, collaboration, reproducibility, or deployment inquiries when the request has a clear purpose and does not conflict with licensing or data-provenance constraints.","Designed and developed independently by Owen Song · open weights · Apache-2.0 · complete local text-to-waveform inference"],"articleImages":[{"sourceUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64812be1856901b0edbcf7ae/KgdBUK01E9utwgcEr18PA.png","alt":"","afterParagraph":0,"url":"/media/articles/cms1i68ky007uroxo9fq348kt/ce282bed8750044c.webp"},{"sourceUrl":"https://huggingface.co/owensong/Inflect-Micro-v2/resolve/main/assets/inflect-v2-repository-hero.png","alt":"Inflect-Micro-v2 release cover","afterParagraph":0,"url":"/media/articles/cms1i68ky007uroxo9fq348kt/4ec23d65da38c191.png"}],"mediaStatus":"ok","articleBodyZh":["：/owensong/Inflect-Micro-v2/blob/main/assets/inflect-v2-repository-hero.png","在不到10M参数下完成本地文本到波形语音合成。固定音色的英语TTS，具有确定性随机种子、长文本处理能力，并支持CPU或CUDA推理。","我独立构建并资助了Inflect v2。如果此版本能够得到真正的受众，我希望能继续开发更广泛的v3版本，可能包括更多语言、声音和稳定性改进。如果这个模型对你有用，在Hugging Face上点个赞真的可以帮助更多人发现它。","：https://huggingface.co/spaces/owensong/Inflect-v2：https://github.com/owenawsong/Inflect：https://huggingface.co/owensong/Inflect-Nano-v2：https://discord.gg/CVJYedvzvp：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/EVALUATION.md","9,356,513 可部署参数 · 37.53 MB FP32 · 24 kHz 单声道输出","Inflect v2在两个规模中使用同一个公共API：Micro优先考虑10M参数以下的质量；Nano优先考虑低于4M参数的占用空间。","这些是保留的文本生成，而非训练音频的重建。每个转录内容都完全按照传递给公共前端的文本显示。","没有单一指标可以衡量TTS质量。Inflect v2分别报告人工偏好、预测自然度、多-ASR可懂度、完整占用空间和运行时间，而不是将它们压缩成一个无法验证的分数。","标题行始终指Inflect-Micro-v2。详细的竞争者结果和协议界限保留在下方可见。","对比集。结果包括KittenTTS Nano：https://huggingface.co/KittenML/kitten-tts-nano-0.8，Piper Low：https://huggingface.co/rhasspy/piper-voices，以及Supertonic 3：https://huggingface.co/Supertone/supertonic-3，这些是已建立的紧凑或本地TTS基线，其可部署权重规模均大于两个Inflect版本。权重大小在软件包级别进行比较，不将单一指标视为整体优越性的证明。","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/human-preference.svg","Inflect-Micro-v2 在最终匿名社区研究中记录了 66.2% 的偏好率（21 胜 · 10 负 · 3 平）。系统被隐藏，左右顺序随机，平局计作半场胜利。这是描述性社区证据，而非正式 MOS。","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/quality-vs-footprint.svg","UTMOS22 测试每个声音使用 500 个相同的未见提示。KittenTTS 和 Piper 是等权重的双声道均值；它们观察到的声音范围显示为须状线。Supertonic 三步法报告在绘图范围下方，而不是对每个其他系统进行拉平处理。","Inflect-Micro-v2：4.395 UTMOS22，95% 自助法置信区间 4.381–4.408。UTMOS22 是学习预测器，而非人工 MOS。","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/asr-consensus.svg","报表分数为每个系统 Qwen3-ASR 和 Nemotron 3.5 语料 WER 的等权重平均值。Whisper 一直被排除在报表之外，因为它在一部分可理解的 Supertonic 八步法剪辑中产生了大量插入性幻觉。它并未删除：完整的三-ASR 证据仍然保留在下方。","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/modern400-three-asr.svg","对于 Inflect-Micro-v2，单项结果为 2.52% Qwen3-ASR、5.45% Nemotron 3.5 和 2.73% Whisper large-v3。前述三模型均值 3.57% 仅作为描述性审计值保留，并未用作报表分数。","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/asr-robustness.svg","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/category-semantic-wer.svg","这些视图是诊断工具，而非额外的排行榜。它们显示了识别器的分歧位置以及哪些提示类别仍会产生可恢复的转录错误。","两个 Inflect 版本在 CPU 上合成速度均高于实时。管理参考运行使用 Hugging Face CPU 升级实例（8 vCPU，32 GB RAM），配备四个框架线程，端到端文本到波形的计时和 100 个固定 Modern400 提示。记录了三次完整运行；第一次缓存构建运行被排除，表格汇总了第二次和第三次运行。","这些是来自公共 PyTorch 运行时的包级别结果，并不是声称 Inflect 是最快的紧凑型 TTS 系统。硬件、前端行为、框架、编译和线程策略都会影响小模型的测量结果。","相同的托管 CPU 和四线程策略被用于较短的比较运行：相同的 50 条提示前缀，重复两次。KittenTTS 和 Piper 的工作量在它们测试的两个语音之间平均分配。","由于 Inflect 使用较大的 100 条提示稳态运行，而比较行使用较短的 50 条提示确认运行，因此此表展示的是部署环境，而非完美匹配的速度排行榜。几个比较模型还使用了优化过的 ONNX 运行时，而上面发布的 Inflect 基准使用的是标准的 PyTorch 运行时。单独发布的 Inflect ONNX 路径尚未替换到那些基准数据中。","：/owensong/Inflect-Micro-v2/blob/main/assets/evidence/model-footprint.svg","共享相同权重的语音变体已合并。Inflect 总计包括集成的波形解码器。","Inflect-Micro-v2 是该系列注重质量的成员。两个模型都使用相同的公共 API 并完成文本到波形的打包。","这使用了 Hub 的版本感知下载器并检索完整的仓库。Git 克隆同样可行，但 hf 下载是普通模型安装推荐路径。","结果是一个 24 kHz 的单声道 float32 波形。长输入在标点感知的边界处分割，逐块合成，并通过控制的停顿拼接。","官方验证的 FP32 导出单独发布为 Inflect-Micro-v2-ONNX：https://huggingface.co/owensong/Inflect-Micro-v2-ONNX。它支持动态长度、CPU/CUDA/DirectML 提供者选择、确定性种子，以及相同的长文本包装器而无需导入 PyTorch：","神经模型被拆分为 duration.onnx 和 decode.onnx；它们共同包含完整的学习文本到波形路径。英语 eSpeak-ng 前端仍然是 CPU 端代码。查看 ONNX 仓库：https://huggingface.co/owensong/Inflect-Micro-v2-ONNX 以获取图表契约、来源、等效性测量、浏览器部署和再导出说明。","Inflect v2 是一个参数高效的 VITS 系列端到端文本到波形生成器，具有英文音素前端、单调对齐、随机潜在合成、残差耦合流和集成化降别名神经波形解码器。","该发布描述了可部署的架构。私有语料库构建和优化细节不属于此开源权重包的一部分。","长文本是按标点符号分段的块，而不是一次无限自动回归处理。块边界会有短暂停顿和边缘渐隐。请参阅 docs/API.md：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/API.md 了解波形协议和并发使用说明。","该发布包含一个固定的合成英文语音。该包不重新分发真实说话者录音语料库，不将该语音宣称为真实人物的身份，并且推理时无需参考音频或外部模型。","此版本以推理为首要目标。目前尚未验证或支持新语音和新语言的适配。新语音将替换固定的说话者，而不是添加可选择的说话者；语言适配还需要重建归一化、音素、符号、嵌入和训练数据。请参阅 docs/DATA_AND_VOICE.md：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/DATA_AND_VOICE.md 和 docs/FINETUNING.md：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/docs/FINETUNING.md。","请勿使用包含的语音冒充真实人物、欺骗听众或制作欺诈内容。在可能导致误导的情况下披露合成语音。用户需遵守适用法律及 Apache-2.0 许可证。","原始 Inflect 代码和权重以 Apache-2.0 发布。捆绑的第三方组件在 THIRD_PARTY_NOTICES.md 中保留其自身声明：https://huggingface.co/owensong/Inflect-Micro-v2/blob/main/THIRD_PARTY_NOTICES.md。release_manifest.json 记录了打包文件大小和 SHA-256 哈希值。","Inflect v2 是开源权重发布版本。可部署权重、推理代码、前端代码、评测提示和发布报告均为公开。训练语料生成流程、私有过滤基础设施和完整优化方案不属于公开包的一部分。","Owen Song 可能会在请求具有明确目的且不违反许可或数据来源限制的情况下，就可信的研究、协作、可重现性或部署相关问题私下分享额外的技术背景。","由 Owen Song 独立设计和开发 · 开源权重 · Apache-2.0 · 完整本地文本到波形推理"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Inflect-Micro-v2 以仅 936 万个参数实现了完整的语音模型功能。 Aioga 将其归入「模型更新」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：模型与研究类动态需要结合能力边界、开放方式、成本、可用性和真实任务表现判断，单项指标领先不等于已经形成稳定采用。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察官方文档、实际可用性、价格变化、开发者反馈和竞品回应。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-07-28T06:29:10.490Z","sourceHash":"79ac29a87e205f27","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["模型更新","Hacker News 热门（buzzing.cc 中文翻译）"],"translations":{"zh-CN":{"title":"Inflect-Micro-v2 发布：仅 936 万参数实现完整语音模型","summary":"Inflect-Micro-v2 以仅 936 万个参数实现了完整的语音模型功能。该模型在 HuggingFace 上发布，参数规模远小于主流语音模型，展示了极低参数量下仍能保持语音生成能力的可行性。","category":"模型更新","source":"huggingface.co","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 发布：仅 936 万参数实现完整语音模型 - Aioga AI资讯","description":"Inflect-Micro-v2 以仅 936 万个参数实现了完整的语音模型功能。该模型在 HuggingFace 上发布，参数规模远小于主流语音模型，展示了极低参数量下仍能保持语音生成能力的可行性。","url":"https://www.aioga.com/news/cms1i68ky007uroxo9fq348kt/"},"en":{"title":"Inflect-Micro-v2 Released: Complete Speech Model Achieved with Only 9.36 Million Parameters","summary":"Inflect-Micro-v2 achieves full speech model functionality with only 9.36 million parameters. The model is released on HuggingFace, and its parameter scale is much smaller than mainstream speech models, demonstrating the feasibility of maintaining speech generation capability with a very low number of parameters.","category":"Models","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 Released: Complete Speech Model Achieved with Only 9.36 Million Parameters - Aioga AI News","description":"Inflect-Micro-v2 achieves full speech model functionality with only 9.36 million parameters. The model is released on HuggingFace, and its parameter scale is much smaller than main...","url":"https://www.aioga.com/en/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:41:53.516Z"},"ja":{"title":"Inflect-Micro-v2 発表：わずか936万パラメータで完全な音声モデルを実現","summary":"Inflect-Micro-v2 はわずか 936 万のパラメータで完全な音声モデル機能を実現しました。このモデルは HuggingFace に公開されており、パラメータ規模は主流の音声モデルよりもはるかに小さく、非常に少ないパラメータ量でも音声生成能力を維持できる可能性を示しています。","category":"モデル更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 発表：わずか936万パラメータで完全な音声モデルを実現 - Aioga AIニュース","description":"Inflect-Micro-v2 はわずか 936 万のパラメータで完全な音声モデル機能を実現しました。このモデルは HuggingFace に公開されており、パラメータ規模は主流の音声モデルよりもはるかに小さく、非常に少ないパラメータ量でも音声生成能力を維持できる可能性を示しています。","url":"https://www.aioga.com/ja/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:42:04.472Z"},"ko":{"title":"Inflect-Micro-v2 발표: 단 936만 개 매개변수로 완전한 음성 모델 구현","summary":"Inflect-Micro-v2는 단 936만 개의 매개변수로 완전한 음성 모델 기능을 실현했습니다. 이 모델은 HuggingFace에 공개되었으며, 매개변수 규모가 주류 음성 모델보다 훨씬 작지만, 매우 적은 매개변수로도 음성 생성 능력을 유지할 수 있음을 보여줍니다.","category":"모델 업데이트","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 발표: 단 936만 개 매개변수로 완전한 음성 모델 구현 - Aioga AI 뉴스","description":"Inflect-Micro-v2는 단 936만 개의 매개변수로 완전한 음성 모델 기능을 실현했습니다. 이 모델은 HuggingFace에 공개되었으며, 매개변수 규모가 주류 음성 모델보다 훨씬 작지만, 매우 적은 매개변수로도 음성 생성 능력을 유지할 수 있음을 보여줍니다.","url":"https://www.aioga.com/ko/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:42:53.099Z"},"es":{"title":"Inflect-Micro-v2 lanzado: modelo de voz completo con solo 9,36 millones de parámetros","summary":"Inflect-Micro-v2 logra funciones completas de modelo de voz con solo 9,36 millones de parámetros. Este modelo se publica en HuggingFace y su tamaño de parámetros es mucho menor que el de los modelos de voz principales, mostrando la viabilidad de mantener la capacidad de generación de voz con un número de parámetros muy bajo.","category":"Modelos","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 lanzado: modelo de voz completo con solo 9,36 millones de parámetros - Aioga Noticias de IA","description":"Inflect-Micro-v2 logra funciones completas de modelo de voz con solo 9,36 millones de parámetros. Este modelo se publica en HuggingFace y su tamaño de parámetros es mucho menor que...","url":"https://www.aioga.com/es/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:42:41.605Z"},"fr":{"title":"Publication de Inflect-Micro-v2 : modèle vocal complet réalisé avec seulement 9,36 millions de paramètres","summary":"Inflect-Micro-v2 réalise les fonctions complètes d'un modèle vocal avec seulement 9,36 millions de paramètres. Ce modèle a été publié sur HuggingFace et sa taille de paramètres est bien inférieure à celle des modèles vocaux principaux, démontrant la faisabilité de maintenir la capacité de génération vocale avec un très petit nombre de paramètres.","category":"Modèles","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Publication de Inflect-Micro-v2 : modèle vocal complet réalisé avec seulement 9,36 millions de paramètres - Aioga Actualités IA","description":"Inflect-Micro-v2 réalise les fonctions complètes d'un modèle vocal avec seulement 9,36 millions de paramètres. Ce modèle a été publié sur HuggingFace et sa taille de paramètres est...","url":"https://www.aioga.com/fr/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:43:34.982Z"},"de":{"title":"Inflect-Micro-v2 veröffentlicht: Vollständiges Sprachmodell mit nur 9,36 Millionen Parametern","summary":"Inflect-Micro-v2 erreicht die vollständige Sprachmodellfunktionalität mit nur 9,36 Millionen Parametern. Das Modell wurde auf HuggingFace veröffentlicht, und die Anzahl der Parameter ist deutlich geringer als bei herkömmlichen Sprachmodellen, was die Machbarkeit zeigt, die Sprachgenerierungsfähigkeit auch bei sehr geringer Parameterzahl aufrechtzuerhalten.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 veröffentlicht: Vollständiges Sprachmodell mit nur 9,36 Millionen Parametern - Aioga KI-News","description":"Inflect-Micro-v2 erreicht die vollständige Sprachmodellfunktionalität mit nur 9,36 Millionen Parametern. Das Modell wurde auf HuggingFace veröffentlicht, und die Anzahl der Paramet...","url":"https://www.aioga.com/de/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:43:38.207Z"},"pt-BR":{"title":"Lançamento do Inflect-Micro-v2: modelo de voz completo com apenas 9,36 milhões de parâmetros","summary":"O Inflect-Micro-v2 alcança funcionalidades completas de modelo de voz com apenas 9,36 milhões de parâmetros. Este modelo foi lançado no HuggingFace, e sua escala de parâmetros é muito menor do que a dos modelos de voz convencionais, demonstrando a viabilidade de manter a capacidade de geração de voz mesmo com um número extremamente baixo de parâmetros.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Lançamento do Inflect-Micro-v2: modelo de voz completo com apenas 9,36 milhões de parâmetros - Aioga Notícias de IA","description":"O Inflect-Micro-v2 alcança funcionalidades completas de modelo de voz com apenas 9,36 milhões de parâmetros. Este modelo foi lançado no HuggingFace, e sua escala de parâmetros é mu...","url":"https://www.aioga.com/pt-BR/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:44:12.684Z"},"ru":{"title":"Inflect-Micro-v2 выпущен: всего 9,36 миллиона параметров для реализации полной голосовой модели","summary":"Inflect-Micro-v2 реализует полные функции речевой модели всего с 9,36 миллионами параметров. Эта модель опубликована на HuggingFace, её размер параметров значительно меньше, чем у основных речевых моделей, демонстрируя возможность сохранять способность к генерации речи при крайне малом количестве параметров.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 выпущен: всего 9,36 миллиона параметров для реализации полной голосовой модели - Aioga Новости ИИ","description":"Inflect-Micro-v2 реализует полные функции речевой модели всего с 9,36 миллионами параметров. Эта модель опубликована на HuggingFace, её размер параметров значительно меньше, чем у...","url":"https://www.aioga.com/ru/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:44:19.838Z"},"ar":{"title":"إصدار Inflect-Micro-v2: نموذج صوت كامل بتحقيق 9.36 مليون معلمة فقط","summary":"تمكن نموذج Inflect-Micro-v2 من تحقيق وظائف نموذج صوتي كامل باستخدام 9.36 مليون معلمة فقط. تم نشر هذا النموذج على HuggingFace، وحجم معلماته أصغر بكثير من النماذج الصوتية الشائعة، مما يوضح إمكانية الحفاظ على القدرة على توليد الصوت حتى مع عدد ضئيل جدًا من المعلمات.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"إصدار Inflect-Micro-v2: نموذج صوت كامل بتحقيق 9.36 مليون معلمة فقط - Aioga أخبار الذكاء الاصطناعي","description":"تمكن نموذج Inflect-Micro-v2 من تحقيق وظائف نموذج صوتي كامل باستخدام 9.36 مليون معلمة فقط. تم نشر هذا النموذج على HuggingFace، وحجم معلماته أصغر بكثير من النماذج الصوتية الشائعة، مم...","url":"https://www.aioga.com/ar/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:45:04.112Z"},"hi":{"title":"Inflect-Micro-v2 जारी: केवल 936 लाख पैरामीटर में पूर्ण ध्वनि मॉडल प्राप्त","summary":"Inflect-Micro-v2 ने केवल 9.36 मिलियन पेरामीटर्स के साथ पूर्ण वॉइस मॉडल कार्यक्षमता हासिल की है। यह मॉडल HuggingFace पर जारी किया गया है, और इसके पेरामीटर का आकार मुख्यधारा के वॉइस मॉडलों की तुलना में काफी छोटा है, जो यह दर्शाता है कि बहुत कम पेरामीटर में भी वॉइस जनरेशन क्षमता बनाए रखना संभव है।","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 जारी: केवल 936 लाख पैरामीटर में पूर्ण ध्वनि मॉडल प्राप्त - Aioga AI समाचार","description":"Inflect-Micro-v2 ने केवल 9.36 मिलियन पेरामीटर्स के साथ पूर्ण वॉइस मॉडल कार्यक्षमता हासिल की है। यह मॉडल HuggingFace पर जारी किया गया है, और इसके पेरामीटर का आकार मुख्यधारा के वॉइस...","url":"https://www.aioga.com/hi/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:45:14.706Z"},"it":{"title":"Rilascio di Inflect-Micro-v2: modello vocale completo realizzato con soli 9,36 milioni di parametri","summary":"Inflect-Micro-v2 realizza tutte le funzionalità di un modello vocale completo con soli 9,36 milioni di parametri. Il modello è pubblicato su HuggingFace e la sua scala di parametri è molto più piccola rispetto ai modelli vocali principali, dimostrando la fattibilità di mantenere la capacità di generare voce anche con un numero di parametri estremamente basso.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Rilascio di Inflect-Micro-v2: modello vocale completo realizzato con soli 9,36 milioni di parametri - Aioga Notizie IA","description":"Inflect-Micro-v2 realizza tutte le funzionalità di un modello vocale completo con soli 9,36 milioni di parametri. Il modello è pubblicato su HuggingFace e la sua scala di parametri...","url":"https://www.aioga.com/it/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:46:08.317Z"},"nl":{"title":"Inflect-Micro-v2 uitgebracht: volledig spraakmodel gerealiseerd met slechts 9,36 miljoen parameters","summary":"Inflect-Micro-v2 bereikt volledige spraakmodelfunctionaliteit met slechts 9,36 miljoen parameters. Dit model is uitgebracht op HuggingFace, met een parameterschaal die veel kleiner is dan die van mainstream spraakmodellen, en laat zien dat het mogelijk is om spraakgeneratiecapaciteit te behouden met een zeer klein aantal parameters.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 uitgebracht: volledig spraakmodel gerealiseerd met slechts 9,36 miljoen parameters - Aioga AI-nieuws","description":"Inflect-Micro-v2 bereikt volledige spraakmodelfunctionaliteit met slechts 9,36 miljoen parameters. Dit model is uitgebracht op HuggingFace, met een parameterschaal die veel kleiner...","url":"https://www.aioga.com/nl/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:45:51.043Z"},"tr":{"title":"Inflect-Micro-v2 Yayınlandı: Sadece 9,36 Milyon Parametre ile Tam Bir Ses Modeli Gerçekleştirildi","summary":"Inflect-Micro-v2, sadece 9,36 milyon parametreyle tam bir ses modeli işlevi gerçekleştirdi. Bu model HuggingFace üzerinde yayımlandı ve parametre boyutu ana akım ses modellerinden çok daha küçüktür, çok düşük parametre sayısında bile ses üretme yeteneğinin sürdürülebilirliğini göstermektedir.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 Yayınlandı: Sadece 9,36 Milyon Parametre ile Tam Bir Ses Modeli Gerçekleştirildi - Aioga AI Haberleri","description":"Inflect-Micro-v2, sadece 9,36 milyon parametreyle tam bir ses modeli işlevi gerçekleştirdi. Bu model HuggingFace üzerinde yayımlandı ve parametre boyutu ana akım ses modellerinden...","url":"https://www.aioga.com/tr/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:46:49.138Z"},"vi":{"title":"Inflect-Micro-v2 ra mắt: Chỉ với 9,36 triệu tham số, đạt được mô hình giọng nói hoàn chỉnh","summary":"Inflect-Micro-v2 chỉ với 9,36 triệu tham số đã thực hiện đầy đủ chức năng mô hình giọng nói. Mô hình này được phát hành trên HuggingFace, quy mô tham số nhỏ hơn nhiều so với các mô hình giọng nói phổ biến, cho thấy khả năng duy trì năng lực tạo giọng nói ngay cả với số lượng tham số rất thấp.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 ra mắt: Chỉ với 9,36 triệu tham số, đạt được mô hình giọng nói hoàn chỉnh - Tin tức AI Aioga","description":"Inflect-Micro-v2 chỉ với 9,36 triệu tham số đã thực hiện đầy đủ chức năng mô hình giọng nói. Mô hình này được phát hành trên HuggingFace, quy mô tham số nhỏ hơn nhiều so với các mô...","url":"https://www.aioga.com/vi/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:46:54.439Z"},"id":{"title":"Inflect-Micro-v2 Dirilis: Hanya 9,36 Juta Parameter Mewujudkan Model Suara Lengkap","summary":"Inflect-Micro-v2 mewujudkan fungsi model suara lengkap dengan hanya 9,36 juta parameter. Model ini dirilis di HuggingFace, dengan skala parameter jauh lebih kecil daripada model suara mainstream, menunjukkan kelayakan kemampuan menghasilkan suara meskipun dengan jumlah parameter yang sangat rendah.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 Dirilis: Hanya 9,36 Juta Parameter Mewujudkan Model Suara Lengkap - Berita AI Aioga","description":"Inflect-Micro-v2 mewujudkan fungsi model suara lengkap dengan hanya 9,36 juta parameter. Model ini dirilis di HuggingFace, dengan skala parameter jauh lebih kecil daripada model su...","url":"https://www.aioga.com/id/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:47:34.589Z"},"th":{"title":"Inflect-Micro-v2 เปิดตัว: ใช้พารามิเตอร์เพียง 9.36 ล้านตัวก็สามารถสร้างโมเดลเสียงครบถ้วน","summary":"Inflect-Micro-v2 สามารถทำงานของโมเดลเสียงได้ครบถ้วนด้วยเพียง 9.36 ล้านพารามิเตอร์ โมเดลนี้ถูกเผยแพร่บน HuggingFace ซึ่งขนาดพารามิเตอร์น้อยกว่าระบบโมเดลเสียงหลักอย่างมาก แสดงให้เห็นถึงความเป็นไปได้ในการรักษาศักยภาพการสร้างเสียงแม้มีพารามิเตอร์จำนวนต่ำ","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 เปิดตัว: ใช้พารามิเตอร์เพียง 9.36 ล้านตัวก็สามารถสร้างโมเดลเสียงครบถ้วน - ข่าว AI Aioga","description":"Inflect-Micro-v2 สามารถทำงานของโมเดลเสียงได้ครบถ้วนด้วยเพียง 9.36 ล้านพารามิเตอร์ โมเดลนี้ถูกเผยแพร่บน HuggingFace ซึ่งขนาดพารามิเตอร์น้อยกว่าระบบโมเดลเสียงหลักอย่างมาก แสดงให้เห็น...","url":"https://www.aioga.com/th/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:47:39.268Z"},"pl":{"title":"Inflect-Micro-v2 wydany: pełny model mowy osiągnięty przy zaledwie 9,36 miliona parametrów","summary":"Inflect-Micro-v2 realizuje pełną funkcjonalność modelu mowy przy zaledwie 9,36 miliona parametrów. Model został opublikowany na HuggingFace, a jego liczba parametrów jest znacznie mniejsza niż w przypadku głównych modeli mowy, co ukazuje wykonalność utrzymania zdolności generowania mowy przy bardzo niskiej liczbie parametrów.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"Inflect-Micro-v2 wydany: pełny model mowy osiągnięty przy zaledwie 9,36 miliona parametrów - Aioga Wiadomości AI","description":"Inflect-Micro-v2 realizuje pełną funkcjonalność modelu mowy przy zaledwie 9,36 miliona parametrów. Model został opublikowany na HuggingFace, a jego liczba parametrów jest znacznie...","url":"https://www.aioga.com/pl/news/cms1i68ky007uroxo9fq348kt/","contentTranslated":true,"sourceHash":"c619c419aa1a3993","translatedAt":"2026-07-27T04:48:17.538Z"}}}}