Industry-Specific Dictionaries and Automatic AI Translation Models
We develop and produce digital, bilingual dictionaries and automatic AI translation assistants (translation and transcription systems), primarily for low-resource languages, as well as the corresponding interfaces (software) for integrating the products into company’s operating systems.
All of our developments are intended solely for the client’s own use – the data is protected against being crawled or scanned by external sources and can therefore also be used offline.
- Preferably in low-resource languages (including Somali, Uzbek, Yiddish, Niger-Congo languages)
- Protected against unauthorized release and the processing of personal data
- Can be used offline
- Can be used in in-house operational systems
- Development of interfaces (software) to integrate the products into existing systems (e.g., intranet)
- Domain-specific customizations: AI-supported data preparation, model architecture and training processes
- Fast development times
- Secure offline development environment
- ISO 9001-certified company
Products and Services
Our translation assistants are tools for representatives of public institutions, companies and translators who work with languages in digital environments as part of their day-to-day work. We develop and produce products for this target group in the fields of language integration and language management for special content-related uses, preferably in low-resource languages.
All of our language models are protected against unauthorized release and the processing of personal information and can therefore also be used offline. The products are integrated into your existing business environment, meaning that no changes need to be made to in-house systems – existing work processes are simply expanded to include the commissioned translation assistance and features.
Digital Dictionaries
Depending on the starting situation, we determine the words to be used from existing and relevant volumes of texts that we either receive from our clients or research independently with our team of translators. This results in a statistically based product that draws from a relevant vocabulary. What’s special about it is that word forms are automatically recognized and assigned, even if a word is searched for with an alternate spelling.
Ongoing monitoring in each language enables us to flexibly adjust the developed dictionaries and keep them dynamic. If needed, we can expand our dictionaries with additional features, including keyword searches or document classifications as just two examples.
Since security for business processes is paramount, all of our dictionaries are protected from unauthorized release and the processing of personal data by external sources. The dictionary systems run exclusively on a company’s in-house network and also work offline – access can additionally be restricted to a desired number of people.
Our digital dictionaries are continuously maintained within the closed system. This enables you to have a custom, individualized dictionary – without giving a single hint about it to the outside world.
SAFIR GmbH delivers the set-up, including the hardware, backend and frontend, if required.
By loading the video, you agree to Vimeo’s privacy policy.
Learn moreAutomatic AI Translation System
In contrast to individual words, this product automatically translates entire phrases and texts. We are currently researching and developing automatic AI translators, primarily for less common languages.
Our AI translation systems can be adapted at any time using individualized training data. The system quickly learns how to translate – that is to say, your preferred in-house communication vocabulary and syntax. You can adapt the vocabulary to the current business case at any time. You train the AI model yourself.
Since security is paramount for business processes, all of our automatic AI translation assistants are protected against unauthorized release and the processing of personal data by external sources. Each model can be run offline, without any connection to the outside world – access can additionally be restricted to a desired number of people.
In the future, SAFIR GmbH delivers the set-up, including the hardware, backend and frontend with the option of easily creating training data sets. If desired, we can also support you in creating the training data sets.
All of our AI models are developed in a closed, offline environment. Furthermore, we can also provide numerous auxiliary programs and processes to help make the operational application and maintenance easier for you. All parts of the system are transparent for you as the client. The technological modules are based on current research results.
Automatic AI Transcription Systems
This product can also automatically transcribe spoken data within your company’s operational infrastructure. In other words, your spoken words become written texts. We are currently researching and developing automatic AI models for transcribing mainly low-resource languages.
Our AI transcription systems can be adapted using individualized training data at any time. The system quickly learns how to transcribe – that is to say, your preferred in-house communication vocabulary and syntax. You can adapt the vocabulary to the current business case at any time. You train the AI model yourself. As an alternative, we can hire language experts to help train the model.
SAFIR GmbH’s AI transcription systems are trained on an ongoing basis to make corrections to audio signals and are robust against incidental noise.
Since security is paramount for business processes, all of our automatic AI transcription systems are protected against unauthorized release and the processing of personal data by external sources. Each model can be run offline, without any connection to the outside world – access can additionally be restricted to a desired number of people.
In the future, SAFIR GmbH delivers the set-up, including the hardware, backend and frontend with the option of easily creating training data sets. If desired, we can also support you in creating the training data sets.
All of our AI models are developed in a closed, offline environment. Furthermore, we can also provide numerous auxiliary programs and processes to help make the operational application and maintenance easier for you. All parts of the system are transparent for you as the client. The technological modules are based on current research results.
Machine translation in times of crisis
We are currently researching the latest AI-based technologies in the project “Machine translation in times of crisis”. The goal is to be able to train domain-specific AI models using our own limited translation data. The project is co-funded by the European Union. The core research is based on synthetic training data generation using innovative AI architectures.
Project Sequence
Research, Analysis & Recruiting
To start, we research the commissioned industry-specific word sets using specialist literature and the internet. We then sort the researched words according to their frequency (corpus analysis), translate them, and develop the corresponding basic morphosyntactic grammar. We work hand in hand with our team of well-trained native speakers who have an affinity for the language required.
Engineering
Developing, testing, documenting: In collaboration with our native speakers, we design the digital translation assistants on the basis of corpus analysis and automate the development of word forms. We develop both the dictionary’s front and back ends and integrate them into your operating systems. We can also expand the software with additional features if so desired. The engineering further includes a continuous monitoring of the use of the respective language.
Project Support
We provide product training in which we teach you how to use the software. Our technical personnel provide direct 1:1 support. Because close customer contact is very important to us, we pay attention to our clients by quickly clarifying questions that arise and promptly reacting your operational needs. Project support is an ongoing process in which we bring ideas to you and take your ideas with us to further refine the dictionary.
ISO 9001
As a company certified to ISO 9001, our quality management focus is on our clients. All work processes are normed and documented.
About ISO 9001
ISO 9001 is the leading international quality management standard around the world. Created by Cert-IT (Certification Authority for IT and Education), it confirms the reliability of a company’s quality assurance (QA) programmes. ISO 9001 stands for quality, security, credibility, trust, and reputation.
Quality Assurance
One of the essential principles of ISO 9001 is a process-oriented approach ensuring that important processes are coordinated with one another. All employees receive the correct training and the right material resources they need to efficiently implement projects. Potential problems are identified and eliminated beforehand, guaranteeing effective internal and external organizational structures.
Quality Goals
As a company certified to ISO 9001, SAFIR GmbH has the following quality goals:
- Offerings are determined by the current state of science and technology
- Customer orders are implemented based on defined budgets
- There is strict adherence to planned delivery times
- We ensure all of our employees are professionally qualified
- The quality of all products is quantified and documented
Client Benefits
As an ISO 9001-certified company, SAFIR GmbH guarantees the following:
- Quality assurance
- High customer satisfaction
- Security
- Good time management
- Efficient project planning
- Audited operational procedures through both internal & external authorities
- In-house documentation & proof of operational procedures
- Long-term & sustainable business development
- Regular risk assessments
SAFIR GmbH
As an ISO 9001-certified company, we have twenty years of experience and expertise in developing and producing digital translation assistants and the associated software for integration into business systems. We specialize in B2B solutions. All of our products are protected from data readout by external sources. SAFIR GmbH delivers innovative customized solutions, security, reliability, and quality products.
Founding
SAFIR GmbH was founded in 1999 by Robert Neumann in Mannheim, Germany. With the advent of computer-aided dictionary generation, the idea of corpus-based language development and associated software in a business context was born.
Expansion
SAFIR GmbH grew in 2010 due to an increase in orders. This is the time when Aria Sheva joined the team. She holds a degree in industrial engineering and is now the company’s CEO.
Berlin
The company relocated to Berlin in 2016 and Jan Neumann became CTO of SAFIR GmbH. Aria Sheva and Jan Neumann also both became shareholders and managing directors, establishing the Berlin office together.
Team
The team at SAFIR GmbH is currently comprised of ten employees in the areas of project management, programming, linguistics and science, and is headed by Managing Directors Aria Sheva and Jan Neumann. We also have a network of up to twenty translators at hand.
Artificial Intelligence
Since 2022, we’ve been investing in the research and development of artificial intelligence. Our goal is to consistently take advantage of new technologies in our proven business concept.
Team
Our team is made up of two managing directors, Jan Neumann and Aria Sheva, two additional full-time employees and up to 25 freelance employees in the areas of project management, programming, linguistics, translation and science. Quality, loyalty and a shared work ethic are what unite us in our daily development and production processes. Our international team is flexible and scalable depending on the workloads.
Jan Neumann
Chief Technology Officer Jan Neumann is the technical director of SAFIR GmbH. He heads the development team and is responsible for information technology, documenting business processes, and quality management within the company.
Aria Sheva-Goldhammer
Chief Executive Officer Aria Sheva-Goldhammer is responsible for project and personnel management, acquisitions and communication at SAFIR GmbH. Ms. Sheva-Goldhammer speaks five languages (including Persian languages, Russian, English) and holds a degree in industrial engineering.
Robert Neumann
Robert Neumann founded the Safir GmbH and today, in his retirement, he works exclusively as an in-house consultant. After studying linguistics, the former director of linguistics data processing at IDS Mannheim has direct access to the scientific and university communities.
Tilmann Cramer
Office Manager Tilmann Cramer oversees our Berlin offices. He coordinates and organises all the day-to-day business and assists with preliminary bookkeeping. He also digitises text materials for our dictionaries.
“We develop dynamic translation tools that don’t require you to pay with your personal or business data.”
Jan Neumann (CTO, SAFIR GmbH)
“We keep our digital translation assistants up to date – even daily, if necessary.”
Robert Neumann (Founder, SAFIR GmbH)
“We develop digital translation assistants for low-resource languages in order to facilitate business communication.”
Aria Sheva-Goldhammer (CEO, SAFIR GmbH)
Clients
Our products are directed toward representatives of public institutions, entrepreneurs in all sectors and translators who require digital and automatic translation tools that can be integrated into company’s existing operational systems, and which are protected against being read by external sources.
Whether companies with business around the globe, public institutions or cultural organizations – our clients include national authorities and the Yiddish Language and Cultural Atlas. We develop and produce specialized, customer-specific translation assistants in primarily low-resource languages. Our working languages are English and German.
Contact
Please use the contact form or write an e-mail to mail@safir-gmbh.de. Thank you.