August 2017 Summaries

The Google Play Developer API has been updated to include new fields for in-app billing information, specifically for Purchases.products and Purchases.subscriptions. Starting August 28, 2017, the API will now return an orderId for both products and subscriptions, with this identifier reflecting the order id in purchases and the most recent recurring order id for subscriptions. Additionally, a new subscription cancelReason value of "2" will indicate that a subscription was canceled because the user changed plans, such as upgrading to a new subscription tier. These updates will be included automatically in the JSON responses to API calls, and developers are advised to review their integrations to ensure compatibility with these changes. For more detailed information, developers can refer to the Purchases.products and Purchases.subscriptions reference pages.

Launchpad Accelerator has expanded its application to include more countries globally, inviting startups from regions such as Africa, Asia, Europe, and Latin America to apply for its equity-free program by October 2, 2017. The program targets technological startups that have moved beyond the ideation stage and are addressing local market challenges, offering them intensive training, mentoring from Google and Silicon Valley experts, Google product credits, and continued support over six months. Selected participants will attend a two-week, all-expense-paid training at Google Developers Launchpad Space in San Francisco, where they will receive guidance from over 20 Google teams and various industry leaders, with the aim of fostering leadership and influence in their local startup ecosystems.

In August 2017, Google announced updates to its AIY Projects, which aim to empower makers of all ages with accessible AI technology. The initiative focuses on creating open and affordable AI kits that simplify human-to-machine interaction, such as the Voice Kit developed in collaboration with Raspberry Pi, which quickly sold out upon release. The AIY team introduced the Speech Commands Dataset, allowing for simple voice interfaces without internet dependency, and is integrating it into the next Voice Kit release. Makers have utilized the kits for various innovative projects, including a retro intercom and a Dalek-voiced assistant. The AIY Projects plan to expand globally and introduce kits with capabilities in vision and motion detection, providing makers with more comprehensive tools for creating intuitive device interfaces. The team invites community input for future developments, encouraging engagement through hackster.io and social media.

Kaldi, a popular open-source speech recognition toolkit, now offers integration with TensorFlow, a move announced by Raziel Alvarez from Google and Yishay Carmiel from IntelligentWire. This integration allows researchers and developers to utilize TensorFlow's deep learning capabilities within Kaldi's speech recognition pipelines, aiming to improve automatic speech recognition (ASR) systems by leveraging deep learning models. The collaboration between Kaldi and TensorFlow promises to enhance the development cycle of ASR systems, reducing the time from model creation to deployment significantly. This is particularly beneficial for companies like IntelligentWire, which focuses on contact center solutions requiring real-time conversation analysis. The integration facilitates the deployment of neural language models, providing improved accuracy and efficiency in transcriptions. It also offers TensorFlow users a robust ASR platform and simplifies the interchange of modules between Kaldi and TensorFlow, fostering closer collaboration between these two open-source communities and paving the way for advancements in speech-based products and research.

Posse, a digital agency, collaborated with the Hamilton musical team to design and launch the official Hamilton app in just three months, utilizing Google's Firebase for backend and infrastructure and Flutter for the front-end UI. The app features a ticket lottery system, user profiles, push notifications, and a custom selfie experience, all powered by Firebase Cloud Functions and Realtime Database, allowing for real-time updates without the need for manual refreshes. Flutter's hot reload capability enabled rapid UI development and iteration, preserving the app state and allowing seamless integration of Hamilton's brand elements across both iOS and Android platforms. The team also created a custom content management system for instant news updates, and used FlutterFire to incorporate Firebase Analytics and Authentication, ultimately recommending serverless architectures and customized UI designs for efficient app development.

TensorFlow Serving 1.0, released in August 2017, marks a significant milestone since its initial open-source launch in February 2016 as a high-performance serving system designed for machine-learned models in production environments. Built from the TensorFlow head, future versions will align with TensorFlow releases, enhancing functionality and API stability for diverse use cases. Initially, it comprised libraries for managing model lifecycles and serving inference requests, later expanding to include a gRPC Model Server binary with a Predict API, deployable on Kubernetes. With over 800 Google projects using it in production, the system is now a stable, robust, and high-performance implementation. The release introduces a prebuilt binary available through apt-get install, simplifying the setup process without compilation, while Docker remains an option for non-Linux installations. The legacy SessionBundle model format is deprecated, with SavedModel, introduced with TensorFlow 1.0, now the supported format, encouraging users to explore the project documentation and tutorials.

Assistant Actions on Google has expanded its availability to British English users, allowing apps built for the Google Assistant to function on Google Home, select Android phones, and iPhones in the UK. Developers are encouraged to optimize their apps for the en-GB audience by selecting from four new TTS voices with British accents and adjusting response text for language differences. Apps like Akinator and Blinkist Minute have already tailored their experiences for these users. Updated developer tools and documentation are available to facilitate the creation, testing, and deployment of apps for the new language setting, with plans for further language expansions in the future. Developers seeking guidance on internationalization can connect with the team via Stackoverflow and Google+.