July 2022 Summaries

Rev AI, a leading speech-to-text company known for its highly accurate speech recognition software, has been integrated into the Eden AI platform and API, expanding its reach to developers and businesses seeking advanced voice applications. Built on over 50,000 hours of human-transcribed content and 12 years of data from Rev's transcription and captioning marketplace, Rev AI's Automatic Speech Recognition (ASR) engine supports 36 languages and offers features like Language Identification, Topic Extraction, and Sentiment Analysis. The service is HIPAA and GDPR compliant, ensuring data security and privacy. With the recent launch of its v2 ASR model, Rev AI has improved its accuracy by over 30% compared to the previous version. The integration with Eden AI allows users to easily implement and scale applications while providing benchmarking tools to assess Rev AI's performance against competitors. Eden AI provides a free tier for testing and pay-as-you-go pricing, making it accessible for a wide range of industries, from business intelligence to real-time captioning.

Computer vision, particularly Face Detection, has gained immense popularity in AI due to its wide-ranging applications across fields like medical imaging, industry, and security. Users can choose between open-source and cloud-based Face Detection engines, each offering distinct advantages and limitations. Open-source engines, available for free and modifiable, provide data privacy but require server hosting and maintenance. Conversely, cloud engines, though paid, offer high performance and server management by the provider, albeit with potential data privacy concerns. A third option is to build a custom Face Detection engine using proprietary data, which requires in-house data science expertise. Eden AI simplifies the process by offering a unified API that allows users to access multiple cloud Face Detection providers, facilitating easy switching and performance comparison without latency issues. This service is particularly beneficial for those seeking to implement Face Detection in JavaScript with minimal effort and time, ensuring GDPR compliance and data security.

Eden AI's front-end interface is designed for simplicity and ease of use, enabling users to compare multiple AI providers through a clear and intuitive user interface built on Vue.js due to its performance and small footprint. The platform employs Bootstrap-Vue to expedite the design process with pre-made components, though customization can be challenging. To guide users in navigating the interface, especially with complex features like the Workflow page, the team utilizes Driver.js for creating tutorials. Furthermore, PapaParse is used for converting CSV files to JSON and vice versa, facilitating data representation. Eden AI's front-end supports AI-powered capabilities such as document processing and language translation by sending data to AI models via API and returning structured JSON results, aggregating the best providers under a single API for seamless comparison and management.

To enhance the user experience on the Eden AI Platform, the team integrated the Cypress tool, an open-source end-to-end testing framework, into their workflow due to its compatibility with JavaScript, which aligns with their front-end development language. Cypress offers automated DOM loading waits, API interaction testing, real-time execution capture, and cross-browser testing capabilities, making it a versatile choice over more complex alternatives like Selenium. The tool's integration into the CI/CD pipeline via Circle CI was straightforward, leveraging Cypress's orb for easy installation and test execution. This integration allows for seamless testing of new features in staging environments and scheduling of daily automated tests, thereby ensuring the platform's performance and reliability for users.

Connexun, an innovative tech startup based in Milan, Italy, has integrated its Sentiment Analysis and Text Summarization API into the Eden AI platform, enhancing its offering of multilingual news and text analysis services. The company's proprietary AI engine, B.I.R.B.A.L., utilizes advanced machine learning techniques, including transformer-based models like GPT-2, to deliver comprehensive NLP capabilities such as summarization, sentiment analysis, and Named Entity Recognition. This integration allows developers easier access to AI services without hosting large models, offering a streamlined experience through Eden AI's platform, which supports composable AI workflows and a flexible pricing model. Connexun aims to empower organizations to transform global news into actionable business insights, catering specifically to media and risk intelligence sectors.

Eden AI utilizes a variety of tools to optimize data collection and analysis for software developers, highlighting the importance of insights derived from data. The company employs Metabase for understanding user information through SQL queries and dashboards, Sentry for tracking and analyzing production errors across applications, and Hotjar for improving user experience by observing user interactions and generating heatmaps. AWS CloudWatch is used to monitor infrastructure metrics such as request numbers and network performance, while Checklyhq provides active monitoring to ensure API responsiveness. These tools collectively enhance the company's ability to maintain efficient operations and improve user experience, with Eden AI also offering a unified API platform that simplifies access to AI capabilities without requiring extensive technical expertise.

NeuralSpace, a Natural Language Processing (NLP) company founded in 2019, has integrated its Translation, Language Detection, and Named Entity Recognition (NER) APIs into Eden AI's platform, with plans to soon add their Speech and Sentiment Analysis APIs. Specializing in local or low-resource languages across Asia, the Middle East, and Africa, NeuralSpace offers a no-code interface that facilitates access to multiple NLP services, including text classification, speaker identification, and speech transcription. Their modular Software-as-a-Service (SaaS) platform allows developers to utilize state-of-the-art NLP models without requiring machine learning expertise and supports custom model training with minimal data input. NeuralSpace's APIs, which can be accessed via REST, are designed for developers in conversational AI and large enterprises needing NLP features across diverse projects. The collaboration with Eden AI aims to provide developers with a wide range of AI services, allowing them to select the most accurate models for their solutions and integrate them seamlessly into AI workflows.

Tabscanner, a leading OCR software specializing in receipt and invoice data extraction, has been integrated into the Eden AI platform and API, enhancing its accessibility and functionality for users. The platform employs advanced machine learning models to accurately extract and classify data from receipts in various languages, demonstrating its versatility by successfully processing over 100 million receipts globally. This technology caters to diverse markets, including expense management, cloud accounting, and market research, and is utilized by prominent companies like Fujitsu, Chevron, and Emirates Airlines. The integration with Eden AI allows Tabscanner to operate seamlessly alongside other AI solutions, providing users with a robust tool for real-time, high-accuracy data processing. Eden AI offers a free tier for testing and flexible pricing options for production use, supporting composable AI workflows that can integrate OCR with other AI capabilities such as NLP and vision.

Speech recognition has become a popular application within artificial intelligence due to its wide range of uses such as in call centers, broadcasting, healthcare, banking, and voice assistants. Key functionalities include Speech-to-Text, Text-to-Speech, Speech Analysis, Speech Diarization, and Speech Translation. Users can choose between open-source engines, which are free and customizable but require self-hosting, and cloud engines, which offer high performance and ease of use but involve data privacy considerations as they are hosted by AI providers. A third option is building a custom engine, which is feasible for companies with data science capabilities. Eden AI offers a unique solution by providing a single API that integrates multiple providers, allowing users to easily compare and switch between them without losing performance, facilitating the use of Speech-to-Text technology in Python.

Speech recognition has gained immense popularity due to its wide range of applications across various sectors like healthcare, banking, and voice assistance. Key functionalities include speech-to-text, text-to-speech, speech analysis, diarization, and translation. Developers seeking to implement speech-to-text capabilities in C# have several options: open-source engines like DeepSpeech and Vosk, which are free and allow customization, or cloud-based solutions from providers such as Google Cloud and Amazon Transcribe, which offer high performance and scalability. While open-source options require self-hosting and maintenance, cloud services handle these aspects at a cost, raising data privacy considerations. A third option is building a custom engine, suitable for companies with data science expertise. Eden AI offers a unified API that simplifies the process by allowing users to access multiple providers with a single token, facilitating comparison and provider switching without performance loss. This flexibility is particularly valuable for developers who want to integrate speech-to-text functionality in under an hour using standardized API endpoints and ready-to-use code examples while ensuring data security and GDPR compliance.

Computer vision, a prominent application within artificial intelligence, is widely utilized across diverse fields such as medical imaging, industry, transport, and security. Face detection, a key functionality, identifies facial positions and other attributes like age and emotions. When selecting a face detection engine, one can choose between open source, cloud-based, or custom-built options, each with distinct advantages and challenges. Open source engines offer flexibility and privacy but require maintenance and server setup, while cloud engines provide ease of use and high performance at the expense of data privacy. A third option is to build a custom engine, beneficial for organizations with data science expertise, ensuring data privacy and tailored performance. Eden AI simplifies the process by offering an API that allows users to access and compare multiple providers, facilitating easy integration and provider switching without compromising latency.

Optical Character Recognition (OCR) is a pivotal component of artificial intelligence and computer vision, facilitating the conversion of various digital text formats, such as PDFs and images, into machine-readable text. Emerging in the 1950s, OCR has evolved to become more intelligent and adaptable, supporting applications across industries like banking, healthcare, and law enforcement. The technology operates in three stages: image pre-processing, character extraction, and post-processing. Users can choose between open-source and cloud-based OCR engines, each offering distinct advantages. Open-source engines, such as pyTesseract and Keras-OCR, are cost-effective and customizable, while cloud engines like those from Google and Microsoft offer robust performance and scalability but at a cost. Eden AI offers a unified API that simplifies the integration of multiple OCR providers, allowing users to switch between them seamlessly based on performance needs, while ensuring data security and compliance with privacy regulations.