Welcome to the AI Daily section! This is your daily guide to exploring the world of artificial intelligence. Every day, we bring you the hottest topics in the AI field, focusing on developers to help you understand technology trends and discover innovative AI product applications.

Discover Fresh AI Products Click to Learn More: https://top.aibase.com/

1、Synthesia Launches New Digital Human Expressive-1 Official Version - Can Understand the Emotions in Your Text

Synthesia has launched Expressive-1 AI Avatars, a technology that can automatically display rich expressions based on text content. This technology enhances video expressiveness and appeal, featuring enhanced expressiveness, synchronized emotional reactions, mimicking human micro-expressions, and body language.

image.png

AiBase Highlights:

✨ Enhanced Expressiveness: Expressive-1 automatically displays adaptive tone, facial expressions, and body language based on text semantics, expressing emotions more naturally.

😊 Synchronized Emotional Reactions: The virtual avatar can accurately display expressions and tones matching the emotional state, improving the accuracy of emotional expression.

🎤 Synchronization of Voice and Lip Movements: Each avatar is equipped with a matching voice and advanced lip-sync technology, ensuring consistency between voice and lip movements.

Details Link: https://top.aibase.com/tool/synthesia

2、iFlytek's Spark Large Model V3.5 Upgrade Introduces Long Text, Long Image, and Long Voice Models

iFlytek has released the Spark Large Model V3.5 upgrade, introducing multi-emotional hyper-realistic synthesis technology, launching long text, long image, long voice models, and the Spark Intelligence Platform, providing powerful technical support for bidding applications and contract applications.

image.png

AiBase Highlights:

🚀 iFlytek releases the Spark Large Model V3.5 upgrade, introducing multi-emotional hyper-realistic synthesis technology.

🔥 Launches the first long text, long image, long voice models, providing more powerful technical support.

💡 The Spark Large Model excels in general long text processing, even surpassing GPT-4 Turbo.

Details Link: https://top.aibase.com/tool/xunfeixinghuorenzhidamoxing

3、iFlytek: Spark Large Model V4.0 Version to be Released on June 27

iFlytek has released the first large model supporting long text, long image, long voice, providing a series of new features including image recognition, contract assistant, intelligent bid evaluation assistant, etc., to achieve more vivid and personalized expressions, solve enterprise implementation issues, and help employees improve work efficiency. Liu Qingfeng announced that the Spark Large Model V4.0 will be officially released on June 27.

AiBase Highlights:

🚀 iFlytek releases the first large model supporting long text, long image, long voice, providing more professional and accurate industry scenario responses.

📝 Launches contract assistant and intelligent bid evaluation assistant, improving contract review efficiency, making bid evaluation more convenient, efficient, and accurate.

🌟 Releases a new intelligent platform, solving large model enterprise implementation issues, creating a personal assistant for employees, helping enterprises liberate productivity.

Details Link: https://top.aibase.com/tool/xunfeixinghuorenzhidamoxing

4、The Stirring Robot is Here! StarDust Smart Launches AI Robot Astribot S1

A domestic AI robot, Astribot S1, which disrupts the household robot field, has been launched, featuring core characteristics such as imitation learning, large model support, and software-hardware collaboration. Its operational performance is excellent, demonstrating multi-task capabilities in home and work scenarios, indicating the wide application potential of AI robots in future life.

image.png

AiBase Highlights:

🤖 Imitation Learning: Astribot S1 can imitate human behavior, perform complex tasks, and demonstrate the agility and flexibility of adults.

🧠 Large Model Support: Connected to a large model test, it is expected to be commercialized within 2024, aiming to establish new AI robot standards.

🔧 Software-Hardware Collaboration: The R&D team has achieved key breakthroughs, enabling S1 to have a smart "brain" and agile "body".

6、OpenVoice V2 Version Released - Can Fine-Tune Voice Styles

The OpenVoice V2 version is an innovative voice cloning technology that can accurately replicate the voice of a reference speaker and generate speech in multiple languages. This version features better audio quality and native multilingual support, integrating MeloTTS technology, supporting free commercial use. The technical approach includes the decoupling design of voice style and language, the basic speaker TTS model with a timbre converter, and training strategies and data processing.

image.png

AiBase Highlights:

✨ Accurate Timbre Cloning: OpenVoice can accurately clone the reference timbre and generate speech in multiple languages.

🔧 Flexible Voice Style Control: Users can finely adjust the emotion, accent, rhythm, pauses, and intonation of the voice, achieving personalized voice output.

🌐 Efficient Computational Performance: OpenVoice significantly reduces computational costs while maintaining high performance.

Official Website: https://research.myshell.ai/open-voice

Project Address: https://top.aibase.com/tool/openvoice

Create Your Own Voice Bot: https://myshell.ai/

7、Intel's First Quarter Performance Shows Strength

Intel's first-quarter revenue reached $12.7 billion, an increase of 9% year-over-year, mainly driven by computing, artificial intelligence, and edge products. Intel has launched a new Gaudi3 AI accelerator, challenging competitors Nvidia and AMD, and making progress in the artificial intelligence field. Intel is accelerating the launch of AI PC products, with an estimated shipment of over 40 million AI PCs by the end of 2024.

AiBase Highlights:

⭐ Intel's first-quarter revenue reached $12.7 billion, an increase of 9%.

⭐ Intel launched a new Gaudi3 AI accelerator, challenging competitors Nvidia and AMD, making progress in the artificial intelligence field.

⭐ Intel is accelerating the launch of AI PC products, with an estimated shipment of over 40 million AI PCs by the end of 2024.

8、Big Tech Engineer Salaries Revealed: OpenAI Engineers Earn Up to $900,000 per Year

In big tech companies, engineers can earn millions of dollars a year, with OpenAI engineers earning up to $900,000. Salaries for engineers at different companies vary, but they are all above a million dollars. Engineers promoted to senior positions can earn several million dollars a year. Talents in the field of artificial intelligence are generously rewarded for their knowledge.

AiBase Highlights:

⭐️ OpenAI engineers earn up to $900,000 per year

⭐️ Google, Apple, Facebook, Microsoft, and other company engineers earn over a million dollars

⭐️ Engineers promoted to senior positions can earn several million dollars a year

9、IntrinsicAnything: Adjusting Image Lighting While Maintaining Object Material

This article introduces a method of learning materials through a generative model, optimizing the process, to improve the accuracy of recovering object materials in images taken under unknown static lighting conditions. Researchers use a model basis of diffuse and specular reflection shading terms, adopting a coarse-to-fine training strategy, to achieve stable and accurate material recovery results.

image.png

AiBase Highlights:

⭐ Learning materials through a generative model, optimizing the process, improving accuracy

⭐ Model based on diffuse and specular reflection shading terms, increasing accuracy

⭐ Adopting a coarse-to-fine training strategy, achieving stable and accurate material recovery results

Details Link: https://top.aibase.com/tool/intrinsicanything

10、Align Your Steps: Maintaining High-Quality Results with Low Step Count Inference

This article introduces a new method called "Align Your Steps," aimed at optimizing the sampling schedule of diffusion models (DMs) in the field of deep learning, enhancing efficiency and quality during the generation process. Through rigorous quantitative experiments, it was found that the optimized schedule significantly improves image quality in image generation benchmarks, and is also applicable to text-to-image and video generation fields.

image.png

AiBase Highlights:

✨ Optimizing sampling schedules to enhance generation model efficiency and quality

🔧 Applicable to various data synthesis benchmark tests, including images, videos, etc.

🚀 Providing user-friendly plug-and-play optimization schedule applications, enhancing stability and quality during the generation process

Details Link: https://top.aibase.com/tool/align-your-steps

11、New ID Preservation Project PuLID: Image Background, Lighting, Style, etc., All Maintain High Consistency

PuLID is a new ID preservation project, committed to enhancing ID preservation effects and minimizing the impact on the original model. Its core advantages include high consistency, versatility, high fidelity, stability, and accuracy, with wide applications. The release of PuLID will promote technological innovation and development, showcasing unique advantages and value. Let's look forward to the release of PuLID and witness its brilliant performance in the technical field.

image.png

AiBase Highlights:

🔍 High Consistency: The background, lighting, layout, and style of the image remain consistent before and after the addition of identity information.

🛠 Versatility: Supports various operations such as style change, IP fusion, accessory modification, attribute editing, and ID mixing, showcasing powerful functionality and effects.

🔒 High Fidelity: Maintains high fidelity while customizing the ID through comparison alignment, providing users with more possibilities and choices.

Details Link: https://top.aibase.com/tool/pulid

12、Physical Education Teacher Arrested for Using AI to Clone Principal's Voice for Retaliation

This article reports the incident of Darion Darrin, a physical education teacher in Baltimore County, Maryland, using AI voice cloning services to frame the principal of Pikesville High School. This incident reveals the risks of AI technology abuse and has sparked social concern about personal information security and privacy protection.

AiBase Highlights:

🔍 AI Clones Principal's Voice: Physical education teacher Darion Darrin arrested for allegedly creating fake recordings.

⚠️ Risk Warning: Abuse of AI voice cloning technology raises social concerns, leading OpenAI to restrict public use of its platform.

🔒 Privacy Protection: Legislators are working to develop laws to protect personal information from being used without permission by tech companies.