jabstone.com

What It Is

JABstone Speech Platform is an AI speech recognition platform for children and adults with speech impairments. The website currently indicates that it is still in beta testing / Coming Soon. It is not a general-purpose transcription tool; instead, it emphasizes “one model per person”: by collecting an individual’s speech samples, it trains a custom recognition model that can understand that person’s unique pronunciation patterns, with a focus on assistive communication use cases.

Core Capabilities

The platform offers a complete workflow from data collection to deployment. Speech Studio generates speech sessions based on real-life scenarios defined by the user, such as communication at school, ordering in a restaurant, or family interactions, helping collect data that better matches individual needs. Speech Editor is used to review, trim, and clean recordings, improving the quality of training samples. Custom AI Models are fine-tuned based on the person’s actual speech patterns, covering atypical pronunciation, apraxia of speech, and similar differences. A key feature is that the model can run in real time on iOS devices; the page explicitly says it requires no internet connection, is fast, and offers better privacy.

Pricing and Support Information

The website does not disclose any free tier, trial policy, official pricing, payment methods, or subscription model, nor does it explain the requirements for applying to the beta. There is also no public information about Chinese-language support, supported languages, recognition accuracy, the amount of training data required, Android/Web support, APIs, or third-party integrations. Its earlier product, JABtalk, was an Android AAC app, but the page states that it is no longer published on Google Play and that only an APK and support page are available for existing users.

Pros and Cons

Its main strength is its highly focused positioning: it addresses the problem that general speech recognition often does not adapt well to people with speech impairments. The workflow of scenario-based data collection plus recording cleanup also fits the logic of personalized model training. On-device offline operation is valuable for schools, homes, communities, and other environments with unstable connectivity, while also helping protect privacy. The limitation is that the product has not yet been officially released, so outsiders cannot evaluate its real-world accuracy, stability, or training cost. It also lacks information on pricing, compliance, data management, and cross-platform support.

Who It’s For and Access from China

It is most relevant for individuals with speech impairments, parents, special education institutions, and rehabilitation or speech therapy professionals who want to follow or try the product. If you need immediately available Chinese speech recognition or a mature AAC tool, you will still need to consider general-purpose speech recognition, mobile assistive communication apps, or local rehabilitation tools. The website provides no evidence regarding access from mainland China, so network connectivity, payment methods, and Chinese model support are all unknown.