Multi-language correction
The product supports auto-correction across many languages, with language-specific models and corpus sizes shown on the site.
NeuroSpell is a deep-learning auto-corrector for spelling and grammar in many languages, with configurable correction levels and on-premise deployment.
NeuroSpell is a deep-learning auto-corrector for spelling and grammar that is presented as a universal tool across many languages. The site positions it for text correction, proofreading, writing aid, OCR error correction, speech-to-text cleanup, and other human-in-the-loop workflows.
The product emphasizes language-specific correction models, configurable correction levels, and support for domain adaptation. It also states that NeuroSpell can be deployed on-premise, and that a bilingual companion, NeuroBiSpell, is available for translation checking and auto-correction.
The product supports auto-correction across many languages, with language-specific models and corpus sizes shown on the site.
The interface exposes correction levels such as Light, Standard, and Extreme, letting users choose between more conservative or more aggressive rewriting behavior.
The site lists capabilities for typo, phonetic, punctuation, fused or split words, complex flexions, elisions, language-specific confusions, non-native errors, and named-entity guidance.
NeuroSpell can be trained on domain vocabulary, sentence patterns, phrasing, and machine-specific errors, which supports specialized workflows.
The product can be deployed on-premise so text does not need to leave the intranet, according to the site copy.
The bilingual NeuroBiSpell demonstrator compares source and target sentences and provides translation-checking support with similarity scoring and auto-corrections.
Use NeuroSpell to proofread multilingual documents where language-specific spelling, punctuation, and grammar issues need automatic correction.
Use the correction-level controls to match the workflow, from conservative checking for polished text to stronger correction for rough drafts or OCR output.
Use the domain-training options when the text contains specialized terminology, recurring phrasing, or machine-generated errors from a specific environment.
Use the on-premise option when text must remain inside an internal network and should not be sent to an external service.
Use NeuroBiSpell when reviewing translations and you need source-target comparison, similarity scoring, and sentence-level correction support.
NeuroSpell is presented as a neural spell and grammar auto-corrector that works in multiple languages. The site also highlights a bilingual variant, NeuroBiSpell, for translation checking.
The French page describes three correction levels: Light, Standard, and Extreme. These map to different tolerance levels depending on writing quality and how aggressively the tool should correct text.
The site says NeuroSpell can be deployed on-premise without sending data outside your intranet. It also notes that texts may be stored for some time on the company’s servers with an IP address for analysis by developers when using the service.
NeuroSpell can be trained on domain-specific vocabulary, sentence structures, phrasing, and machine-specific errors. The bilingual page also says it can be trained on in-domain corpora and style guides.
The source mentions GitHub, Twitter, Discord, and LinkedIn as places to report bugs or chat with the team, but it does not provide a detailed integration list for third-party applications.