亞太和非洲語言市場消費者設備全球語音-AI公司語音建設認可

這是Lifewood數據技術在多語種語音+LLM中的一項案例研究,這是爲全球語音-AI公司在亞太地區和全中國非洲語言市場建立消費者設備語音識別服務而進行的一項參與。 14,000 發言時數 11 語文和 8 國家,切換字數錯誤率 92%

以15種語言工作的語音助理 在所有需要成長的地方都失敗了

客戶端的語音助理在15種主要語言上工作良好,但在新興市場卻失敗了,數億潛在用戶在數字語音數據最少的情況下會說語言,目標語言不存在現成語音數據集,客戶端無法在道德上大規模刮掉語音數據. 需要在整個地理領域進行收集工作,因爲客戶沒有業務足跡。 造成這種收集工作的制約因素是代表性問題。 發言者必須跨越不同的年齡、性別、口音和錄音環境,因此,完成的語音AI將服務於全體人口,而不是城市子集。 在一個地方,一個方便收集的數據集複製了該地的人口統計,而不管其增長多大——這正是失敗模式的衆包平臺產生的,因爲它們向城市、受過教育的高資源語言人口搖擺不定,系統地忽略了方言和記錄了變化。

在8個國家開展外勤業務,在社區內而不是在網上徵聘

Lifewood activated its field operations in 8 African and Southeast Asian countries, recruiting 6,200+ native speakers across rural and urban communities. Contributors were ethically compensated above local fair-wage benchmarks with full informed-consent documentation. Collection covered scripted prompts, spontaneous conversation, and domain-specific commands across commerce, navigation and media, tailored to the client's assistant use cases. 1. In-community recruitment. Contributors are recruited from the language community itself through field operations, not from crowdsourcing platforms — which is what makes authentic dialect, register and accent coverage possible at all. 2. Demographic balancing by design. Speaker panels were balanced across age, gender, accent and recording environment as a recruitment specification rather than as a post-hoc filter. 3. Environmental diversity sampling. Recording conditions were varied deliberately so the model learns the speech rather than the studio. 4. Phonetic transcription and quality control. Quality control included phonetic transcription, speaker demographic balancing and environmental noise diversity sampling, under Lifewood's 95%+ accuracy SLA and dual-layer human-in-the-loop review. Lifewood's language coverage behind this programme spans 50+ languages reaching more than 90% of the global population, including Swahili, Wolof, Hausa, Amharic, Tigrinya, Yoruba, Zulu, Shona, Lingala and Somali across Africa, and Tagalog, Cebuano, Ilokano, Waray, Khmer, Tok Pisin, Tetum, Fijian and Samoan across Southeast Asia and the Pacific.

字差率下降92%,9個月內運送11種新市場語言

方案已交付 14,000 發言時數 11 來自 6,200+ 獨有的演講者 8 。在客戶端,目標語言的字數錯誤率下降 92% 對照方案前基線, 11 新的市場語言在方案啓動後9個月內在助理人員中投入使用,而工作基礎爲: 15 訂婚前主要語言。

相關 Lifewood 服務

• 低資源語音數據收集
• 多語種培訓數據收集
• 企業 LLM 培訓數據