Apple quietly shipped everything you need to build a real-time translator — so I built one
A developer created Wakaru, a macOS app for live audio translation, leveraging new on-device speech, translation, and LLM APIs introduced in macOS 14 (Sonoma). The application performs all processing locally, ensuring privacy and real-time performance.
- Apple's macOS 14 (Sonoma) includes robust on-device APIs for speech recognition, translation, and large language models.
- Developers can now build powerful, privacy-focused AI applications that run entirely on the user's device, without cloud dependency.
- The creation of a real-time audio translator demonstrates the practical capabilities and potential of Apple's local machine learning framework.
- On-device AI processing offers benefits in privacy, latency, and offline functionality for end-users.
A developer has successfully built Wakaru, a macOS application that provides live, translated subtitles for any audio playing on a Mac. This innovative tool operates entirely on-device, utilizing the new speech, translation, and large language model (LLM) APIs that Apple integrated into macOS 14, also known as Sonoma.
The project highlights Apple's quiet but significant advancements in making powerful machine learning capabilities accessible to developers for local execution. By processing all audio and translation tasks directly on the user's Mac, Wakaru ensures enhanced privacy and eliminates the need for cloud-based services, which often come with latency and data security concerns.
This demonstration serves as a practical example of how developers can leverage Apple's framework to create sophisticated AI-powered applications that benefit from the performance and security of on-device processing. It underscores a growing trend towards local AI, enabling new categories of software that prioritize user data and offline functionality.
Highlights new powerful on-device AI APIs in macOS for building privacy-preserving applications.
Shows potential for new privacy-centric AI products and features on Apple platforms, reducing reliance on cloud services.
Demonstrates the increasing capability of on-device AI for practical, everyday applications like real-time translation.
AI ToolsOpenAI launches ChatGPT desktop app for Linux
AI ToolsMAI-Code-1-Flash Highlights Microsoft’s Lower-Cost Approach to Coding AI
AI ToolsI lost my best AI prompt after 40 tweaks. So I built a tiny git for prompts.
AI ToolsWhy your agent's tool list keeps drifting away from your database
AI ToolsEvery Coding Agent Session Starts by Rediscovering Your Repository
USC Scientists Are Using Quantum Computing to Rethink Cancer Detection - USC Viterbi School of Engineering
USC scientists are leveraging quantum computing to revolutionize cancer detection methods.
FundingAccel closes oversubscribed $550M India fund within weeks, 19 months after its last
Accel closed a $550 million India fund in weeks, 19 months after its last one, with over half of the prior $650 million still unspent.
SecurityChrome adopts what may be the best protection yet against account takeovers
Google Chrome now uses device-bound session credentials to prevent account takeovers, a growing threat in online security.
Meta can't stop states' $1.4 trillion lawsuit from going to trial
A federal judge ruled that Meta cannot dismiss a $1.4 trillion lawsuit filed by multiple states using Section 230 as a defense, allowing the case to proceed to trial.
Air Force explores AI tools to predict aircraft failures, strengthen sustainment - DefenseScoop
The US Air Force is exploring the use of AI tools to predict aircraft failures and improve sustainment. This move aims to reduce maintenance costs and increase aircraft availability.
BusinessGemini becomes Google's fastest-growing product ever as it hits 1B users
Google announced that its AI assistant Gemini has reached 1 billion users, making it the fastest-growing product in the company's history.