The Twopan Portable SSD 2TB is a good external storage option intended for your iPhone and iPad, but its retail cost is prohibitive, so wait for a good sale.Twopan Portable SSDBeating the storage limits of an iPhone or iPad can be done in two ways. You can use cloud storage, or your can attach an external storage device.All the second option requires is an external drive with a USB-C connection that can plug into an iPhone, iPad, or Mac. However, typical external drives take up space, and aren't great for carrying around. Continue Reading on AppleInsider | Discuss on our Forums

Among the backlash Nintendo has been facing for its pricing decisions regarding the upcoming Switch 2 as well as its launch titles is also the fact that Nintendo Switch 2 Welcome Tour is not a pack-in title. Rather, players will need to spend $9.99 to play the game. In the meantime, former Nintendo of America president Reggie Fils-Aime spoke about the decision to offer Wii Sports as a pack-in title for the immensely popular Wii. In a conversation with IGN during its Nintendo Voice Chat podcast, Fils-Aime recounted a story about an argument he had with Shigeru Miyamoto surrounding whether or not Nintendo should offer Wii Sports as a pack-in title for the console. According to the story, Miyamoto was quite against including the title as a free pack-in with the console. “It’s an understatement to say that Mr. Miyamoto pushed back,” said Fils-Aime. “During the development of the Wii and the Wii Sports software, you know again the goal was, we wanted the system to be attractive not only to the most active players, but to new players, and to bring them into the video gaming experience. And we saw Wii Sports as the way to be able to do that because each of these sports made great use of the [Wiimote]. Each sport was known by a global population… and my recommendation was that we pack Wii Sports in with the hardware.” “And literally, when I first made this suggestion, Mr. Miyamoto said ‘Reggie, Nintendo does not give away software for free. You don’t understand how hard our developers work to create compelling content like this,'” continued Fils-Aime. “Now, fortunately, because my SNES came packed in with software, I knew that in fact the company has given away software in the past. But it was against a very strategic objective. And that was my commentary back to Mr. Miyamoto. That, by including Wii Sports, we would immediately have value in the hardware as the consumers opening it up and setting it up. But more importantly, we would have a touchstone piece of software that so many consumers would experience” “It’s what led to Wii Sports being played in bars, being played on cruise ships, being played in retirement homes. It became this universal experience. But he was not happy when I first made the suggestion, to the point where, on a subsequent visit to Kyoto and the headquarters, the team showed me software that would become Wii Play. Those themselves were fun as well, but they didn’t have the connective tissue that Wii Sports had with all of the experiences. So then I proceeded to piss Mr. Miyamoto off again by suggesting that we bundle Wii Play with the Wii Remote.” “In the end, in the Americas and in Europe, Wii Sports was packed in with the Wii. It was not in Japan, which created a bit of a test market, and it was obvious that, in the markets where Wii Sports was packed in that the Wii became much more of a phenomenon. Wii Sports itself became much more of a phenomenon. And we did pack the Wiimote with Wii Play, and I think it became the fifth best selling piece of software in the history of Wii.” Incidentally, Nintendo hadn’t released any pack-in games with the Switch when it launched back in 2017. Rather, the company has been focused on selling even titles like 1-2-Switch separately, sticking to the philosophy that Miyamoto had told Fils-Aime about in the story. The story of Wii Sports pack in …https://t.co/LhflSFWaL3— Reggie Fils-Aimé (@Reggie) April 9, 2025

The Jewish Cemetery is an oil-on-canvas painting by the Dutch landscape painter Jacob van Ruisdael. Painted in 1654 or 1655, it is an allegorical landscape painting suggesting ideas of hope and death, while also being based on Beth Haim, a cemetery located on Amsterdam's southern outskirts, at the town of Ouderkerk aan de Amstel. Beth Haim is a resting place for some prominent figures among Amsterdam's large Jewish Portuguese community in the 17th century. Ruisdael presents the cemetery as a landscape variant of a vanitas painting, employing deserted tombs, ravaged churches, stormy clouds, dead trees, changing skies, and flowing water to symbolize death and the transience of all earthly things. The known provenance for the painting dates back only to 1739 and its original owner is not documented; since 1926, it has been owned by the Detroit Institute of Arts. Painting credit: Jacob van Ruisdael Recently featured: Gatekeeper Florence Price Blue-ice area Archive More featured pictures

April 11 Mary II and William III 1689 – William III and Mary II (both pictured) were crowned joint sovereigns of England in a ceremony at Westminster Abbey. 1809 – Napoleonic Wars: A hastily assembled Royal Navy fleet launched an assault against the main strength of the French Atlantic Fleet; an incomplete victory led to political turmoil in Britain. 1951 – U.S. president Harry S. Truman relieved General of the Army Douglas MacArthur of his commands for making public statements about the Korean War that contradicted the administration's policies. 2001 – In a FIFA World Cup qualifying match, Australia defeated American Samoa 31–0, the largest margin of victory recorded in international football. Romanos III Argyros (d. 1034)Ewelina Hańska (d. 1882)Trevor Linden (b. 1970) More anniversaries: April 10 April 11 April 12 Archive By email List of days of the year About

Hey everyone, Just wanted to say hi and share a 3D environment piece I’ve been working on. It was a fun project and a solid opportunity to dive into some new techniques. Full piece (with video!) here: ArtStation - Land…

See the Face of a Royal Woman Who Lived in Greece 3,500 Years Ago Created by digital artist Juanjo Ortega G., the digital reconstruction depicts a woman who died in her mid-30s during the late Bronze Age Digital artist Juanjo Ortega G. created an image of the woman's face based on a clay model created in the 1980s. Juanjo Ortega G. Roughly 3,500 years ago, a woman was buried in a royal cemetery in present-day Greece. Now, we can imagine what she might have looked like during the late Bronze Age. Digital artist Juanjo Ortega G. has created an image of the woman’s face that allows historians to “peer back into the eyes of the past,” says Emily Hauser, who commissioned the artwork, to the Observer’s Dalya Alberge. Hauser, a historian and classics scholar at the University of Exeter, writes about the woman in her forthcoming book, Penelope’s Bones: A New History of Homer’s World Through the Women Written Out of It. In the book, which comes out later this year, Hauser tells the often-overlooked stories of women in ancient Greece—including the mysterious royal woman. “Too many of the faces we see peering back at us from the ancient world are those of men,” Hauser writes in a social media post. While working on the book, she hired Ortega G. to create a digital model of the woman’s face, based on a clay reconstruction made by researchers at Manchester University in the 1980s. The artist also drew on wall paintings from the time, particularly those from Santorini, where researchers discovered “a striking image of a woman with red-gold hair and blue eyes—which he used as his inspiration here,” Hauser tells Smithsonian magazine. The woman’s remains were discovered in the 1950s at the Mycenae archaeological site in Greece. In Greek mythology—including Homer’s epic poems—Mycenae was ruled by Agamemnon, the king who led the attack on Troy during the Trojan War. “For the first time, we are looking into the face of a woman from a kingdom associated with Helen of Troy—Helen’s sister, Clytemnestra, was queen of Mycenae in legend—and from where the poet Homer imagined the Greeks of the Trojan War setting out,” Hauser tells the Observer. Though the woman died “several hundred years before the supposed date of the Trojan war,” Hauser adds, “such digital reconstructions persuade us that these were real people.” An analysis revealed the woman was in her mid-30s at the time of her death and was suffering from arthritis in her hands, probably due to a repetitive activity like weaving. She died during the late Bronze Age, between the 16th and 17th centuries B.C.E. The woman was found buried with a death mask and a warrior kit that contained various weapons, including three swords. But because she was interred next to a man, researchers long thought those artifacts belonged to him, not her. They also suspected she was married to the man. But DNA analysis later revealed that the man was her brother, not her husband. Researchers now suspect the grave goods may have belonged to the woman, who was “buried there by virtue of her birth, not her marriage,” Hauser tells the Observer. “The remains from the ancient world are telling us a different story from the one we always thought,” she writes in the Conversation. “A woman didn’t have to be a wife to make a difference.” More broadly, the woman is a “fascinating case study” of the various ways archaeologists have misread the women of the ancient past, Hauser tells Smithsonian magazine. “She’s doing a lot to tell us more about women in the Late Bronze Age world—and to push against the biases not just of her own ancient world, but of the later archaeologists who uncovered her, too,” she adds. “And that’s a fascinating story to tell in how it opens up a previously unknown side of the prehistoric Greek world.” Get the latest stories in your inbox every weekday.

DeepCoder-14B competes with frontier models like o3 and o1—and the weights, code, and optimization platform are open source.Read More

Microsoft is starting to gradually roll out a preview of Recall, its feature that captures screenshots of what you do on a Copilot Plus PC to find again later, to Windows Insiders, according to a blog post published Thursday. This new rollout could indicate that Microsoft is finally getting close to launching Recall more widely. Microsoft originally intended to launch Recall alongside Copilot Plus PCs last June, but the feature was delayed following concerns raised by security experts. The company then planned to launch it in October, but that got pushed as well so that the company could deliver “a secure and trusted experience.” The company did release a preview of Recall in November to Windows Insiders in the Dev Channel for Qualcomm Copilot Plus PCs and made a preview available to Intel- and AMD-powered Copilot Plus PC shortly after. And after a couple weeks of testing, my colleague Tom Warren said that Recall is “creepy, clever, and compelling.” In Thursday’s blog post, Microsoft spells out that you have to opt in to saving snapshots with Recall, and you can pause saving them “at any time.”

In today’s enterprise landscape—especially in insurance and customer support —voice and audio data are more than just recordings; they’re valuable touchpoints that can transform operations and customer experiences. With AI audio processing, organizations can automate transcriptions with remarkable accuracy, surface critical insights from conversations, and power natural, engaging voice interactions. By utilizing these capabilities, businesses can boost efficiency, uphold compliance standards, and build deeper connections with customers, all while meeting the high expectations of these demanding industries. Boson AI introduces Higgs Audio Understanding and Higgs Audio Generation, two robust solutions that empower you to develop custom AI agents for a wide range of audio applications. Higgs Audio Understanding focuses on listening and contextual comprehension. Higgs Audio Generation excels in expressive speech synthesis. Both solutions are currently optimized for English, with support for additional languages on the way. They enable AI interactions that closely resemble natural human conversation. Enterprises can leverage these tools to power real-world audio applications. Higgs Audio Understanding: Listening Beyond Words   Higgs Audio Understanding is Boson AI’s advanced solution for audio comprehension. It surpasses traditional speech-to-text systems by capturing context, speaker traits, emotions, and intent. The model deeply integrates audio processing with a large language model (LLM), converting audio inputs into rich contextual embeddings, including speech tone, background sounds, and speaker identities. The model achieves nuanced interpretation by processing these alongside text tokens, essential for tasks such as meeting transcription, contact center analytics, and media archiving. A key strength is its chain-of-thought audio reasoning capability. This allows the model to analyze audio in a structured, step-by-step manner, solving complex tasks like counting word occurrences, interpreting humor from tone, or applying external knowledge to audio contexts in real time. Tests show Higgs Audio Understanding leads standard speech recognition benchmarks (e.g., Common Voice for English) and outperforms competitors like Qwen-Audio, Gemini, and GPT-4o-audio in holistic audio reasoning evaluations, achieving top scores (60.3 average on AirBench Foundation) with its reasoning enhancements. This real-time, contextual comprehension can give enterprises unparalleled audio data insights. Higgs Audio Generation: Speaking with Human-Like Nuance   Higgs Audio Generation, Boson AI’s advanced speech synthesis model, enables AI to produce highly expressive, human-like speech essential for virtual assistants, automated services, and customer interactions. Unlike traditional text-to-speech (TTS) systems that often sound robotic, Higgs Audio Generation leverages an LLM at its core, enabling nuanced comprehension and expressive output closely aligned with textual context and intended emotions. Boson AI addresses common limitations of legacy TTS, such as monotone delivery, emotional flatness, incorrect pronunciation of unfamiliar terms, and difficulty handling multi-speaker interactions, by incorporating deep contextual understanding into speech generation. The unique capabilities of Higgs Audio Generation include: Emotionally Nuanced Speech: It naturally adjusts tone and emotion based on textual context, creating more engaging and context-appropriate interactions. Multi-Speaker Dialogue Generation: This technology simultaneously generates distinct, realistic voices for multi-character conversations, as Boson AI’s Magic Broom Shop demo Accurate Pronunciation and Accent Adaptation: Precisely pronounces uncommon names, foreign words, and technical jargon, adapting speech dynamically for global and diverse scenarios. Real-Time Generation with Contextual Reasoning: This technology produces coherent, real-time speech outputs responsive to conversational shifts, suitable for interactive applications like customer support chatbots or live voice assistants. Benchmark results confirm Higgs Audio’s superiority over top competitors, including CosyVoice2, Qwen2.5-omni, and ElevenLabs. In standard tests like SeedTTS and the Emotional Speech Dataset (ESD), Higgs Audio achieved significantly higher emotional accuracy, while being competitive or superior in word error rate (~1.5–2%). This performance demonstrates Higgs Audio’s ability to deliver unmatched clarity, expressiveness, and realism, setting a new benchmark for audio generation. Under the Hood: LLMs, Audio Tokenizers, and In‑Context Learning   Boson AI’s Higgs Audio models leverage advanced research, combining LLMs with innovative audio processing techniques. At their core, these models utilize pretrained LLMs, extending their robust language understanding, contextual awareness, and reasoning abilities to audio tasks. Boson AI achieves this integration by training LLMs end-to-end on extensive paired text–audio datasets, enabling semantic comprehension of spoken content and acoustic nuances. Boson AI’s custom audio tokenizer is a critical element that efficiently compresses raw audio into discrete tokens using residual vector quantization (RVQ). This preserves linguistic information and subtle acoustic details (tone, timbre) while balancing token granularity for optimal speed and quality. These audio tokens seamlessly feed into the LLM alongside text, allowing simultaneous processing of audio and textual contexts. Also, Higgs Audio incorporates in-context learning, enabling models to adapt quickly without retraining. With simple prompts, such as brief reference audio samples, Higgs Audio Generation can instantly perform zero-shot voice cloning, matching speaking styles. Similarly, Higgs Audio Understanding rapidly customizes outputs (e.g., speaker labeling or domain-specific terminology) with minimal prompting. Boson AI’s approach integrates transformer-based architectures, multimodal learning, and Chain-of-Thought (CoT) reasoning, enhancing interpretability and accuracy in audio comprehension and generation tasks. By combining LLM’s strengths with sophisticated audio tokenization and flexible prompting, Higgs Audio delivers unprecedented performance, speed, and adaptability, significantly surpassing traditional audio AI solutions. Benchmark Performance: Outpacing Industry Leaders   Boson AI extensively benchmarked Higgs Audio, confirming its competitive leadership in audio understanding and generation compared to top industry models. In audio understanding, Higgs Audio matched or surpassed models like OpenAI’s GPT-4o-audio and Gemini-2.0 Flash. It delivered top-tier speech recognition accuracy, achieving state-of-the-art Mozilla Common Voice (English) results, robust performance on challenging tasks like Chinese speech recognition, and strong results on benchmarks such as LibriSpeech and FLEURS.   However, Higgs Audio Understanding truly differentiates itself in complex audio reasoning tasks. On comprehensive tests like the AirBench Foundation and MMAU benchmarks, Higgs outperformed Alibaba’s Qwen-Audio, GPT-4o-audio, and Gemini models, scoring an average of 59.45, which improved to above 60 with CoT reasoning. This demonstrates the model’s superior capability to understand nuanced audio scenarios and dialogues with background noise and interpret audio contexts logically and insightfully. On the audio generation side, Higgs Audio was evaluated against specialized TTS models, including ElevenLabs, Qwen 2.5-Omni, and CosyVoice2. Higgs Audio consistently led or closely matched competitors on key benchmarks: Seed-TTS Eval: Higgs Audio achieved the lowest Word Error Rate (WER), indicating highly intelligible speech, and demonstrated the highest similarity to reference voices. In comparison, ElevenLabs had slightly lower intelligibility but notably weaker voice similarity. Emotional Speech Dataset (ESD): Higgs Audio achieved the highest emotional similarity scores (over 80 versus mid-60s for ElevenLabs), excelling in emotionally nuanced speech generation. Boson AI also introduced the “EmergentTTS-Eval,” using advanced audio-understanding models (even competitors like Gemini 2.0) as evaluators. Higgs Audio was consistently preferred over ElevenLabs in complex scenarios involving emotional expression, pronunciation accuracy, and nuanced intonation. Overall, benchmarks clearly show Higgs Audio’s comprehensive advantage, ensuring users adopting Boson AI’s models gain superior audio quality and insightful understanding capabilities. Enterprise Deployment and Use Case: Bringing Higgs Audio to Business   Higgs Audio Understanding and Generation function on a unified platform, enabling end-to-end voice AI pipelines that listen, reason, and respond, all in real time. Customer Support: At a company like Chubb, a virtual claims agent powered by Higgs Audio can transcribe customer calls with high accuracy, detect stress or urgency, and identify key claim details. It separates speakers automatically and interprets context (e.g., recognizing a car accident scenario). Higgs Audio Generation responds in an empathetic, natural voice, even adapting to the caller’s accent. This improves resolution speed, reduces staff workload, and boosts customer satisfaction. Media & Training Content: Enterprises producing e-learning or training materials can use Higgs Audio Generation to create multi-voice, multilingual narrations without hiring voice actors. Higgs Audio Understanding ensures quality control by verifying script adherence and emotional tone. Teams can also transcribe and analyze meetings for speaker sentiment and key takeaways, streamlining internal knowledge management. Compliance & Analytics: In regulated industries, Higgs Audio Understanding can monitor conversations for compliance by recognizing intent beyond keywords. It detects deviations from approved scripts, flags sensitive disclosures, and surfaces customer trends or pain points over thousands of calls, enabling proactive insights and regulatory adherence. Boson AI offers flexible deployment, API, cloud, on-premise or licensing, with models that adapt via prompt-based customization. Enterprises can tailor outputs to domain-specific terms or workflows using in-context learning, building intelligent voice agents that match internal vocabulary and tone. From multilingual chatbots to automated meeting summaries, Higgs Audio delivers conversational AI that feels truly human, raising the quality and capability of enterprise voice applications. Future Outlook and Strategic Takeaways   Boson AI’s roadmap for Higgs Audio indicates a strong future pipeline of features to deepen audio understanding and generation. A key upcoming capability is multi-voice cloning, allowing the model to learn multiple voice profiles from short samples and generate natural conversations between the speakers. This will enable use cases like AI-powered cast recordings or consistent virtual voices across customer touchpoints. This goes beyond current one-speaker cloning, with Boson AI’s TTS demo already hinting at its arrival. Another development is explicit control over style and emotion. While the current model infers emotion from context, future versions may allow users to specify parameters like “cheerful” or “formal,” enhancing brand consistency and user experience. The Smart Voice feature previewed in Boson AI’s demos suggests an intelligent voice-selection system tailored to script tone and intent. On the understanding side, future updates may enhance comprehension with features like long-form conversation summarization, deeper reasoning via expanded chain-of-thought capabilities, and real-time streaming support. These advancements could enable applications like live analytics for support calls or AI-driven meeting insights. Strategically, Boson AI positions Higgs Audio as a unified enterprise audio AI solution. By adopting Higgs Audio, companies can access the frontier of voice AI with tools that understand, reason, and speak with human-level nuance.  Its dual strength in understanding and generation, built on shared infrastructure, allows seamless integration and continuous improvement. Enterprises can benefit from a consistent platform where models evolve together, one that adapts easily and stays ahead of the curve. Boson AI offers a future-proof foundation for enterprise innovation in a world increasingly shaped by audio interfaces. Sources Thanks to the Boson AI team for the thought leadership/ Resources for this article. Boson AI team has financially supported us for this content/article. Asif RazzaqWebsite |  + postsBioAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.Asif Razzaqhttps://www.marktechpost.com/author/6flvq/Interview with Hamza Tahir: Co-founder and CTO of ZenMLAsif Razzaqhttps://www.marktechpost.com/author/6flvq/OpenAI Open Sources BrowseComp: A New Benchmark for Measuring the Ability for AI Agents to Browse the WebAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Google Introduces Agent2Agent (A2A): A New Open Protocol that Allows AI Agents Securely Collaborate Across Ecosystems Regardless of Framework or VendorAsif Razzaqhttps://www.marktechpost.com/author/6flvq/OpenAI Introduces the Evals API: Streamlined Model Evaluation for Developers

Last updated April 10, 2025: Checked for new Five Nights TD codes!Need some extra units to help vanquish the nightmares lurking in Five Nights TD? We've rounded up all of the Roblox strategy game's active codes, so you can earn some tokens, souls, and other useful goodies.Active Five Nights TD Codes (April 2025)Below, you'll find all of the currently active and working Five Nights TD codes that you can redeem for free rewards in April 2025:UPDATE42 - 2000 Space Tokens (NEW)APRILFOOLS - 2000 Tokens (NEW) BEAR5 - 15 Souls (NEW)SPACEENDLESS - 15 Souls (NEW) SPACEAPEX - 3000 space tokens (NEW)NEBULA - 2,000 Space TokensFREDDYINSPACE - 15 Souls ALIENHELPY - 15 Souls SEASON7 - 1x Exclusive Pack 7 RUIN - 30 SoulsPIRATEFOXY - 15 Souls SECURITYBREACH - 1x Exclusive Pack 7 FIVENIGHTSTD - 1,000 TokensExpired Five Nights TD CodesBelow, you'll find a list of expired Five Nights TD codes that can no longer be redeemed as of April 2025:UPDATE41UPDATE40AUCTIONSPLAZAREVAMP GREEKENDLESS OLYMPUS FOXSEIDON ZEUSFREDEUSGLAMROCKSEASON6TITLES VALENTINEUNITMANAGERUPDATE35 UPDATE34ARG2UPDATE33CHRISTMASSANTASHOPGOLIATHEVONEWQUESTSUPDATE32UPDATE30ENDLESS7HAPPYNEWYEARUPDATE29PIZZASIMSEASON5CHRISTMASENDLESSUPDATE28UPDATE27CALENDERGIFT UPDATE26RANKED2V2470MVISITSSTEAMPUNKENDLESSSTEAMPUNKUPDATE25RANKEDVERSUSUPDATE24KRONOSUPDATE23NEWLOBBYSPROCKETPOTIONSHow to Redeem Five Nights TD CodesBelow, you'll find all of the currently active and working Five Nights TD codesTo redeem Five Nights TD codes, boot up the Roblox Experience and, once you’re in, follow the steps below:Look at the icons running along the bottom of the screen. You’ll see three icons.Click the gear icon to access the settings menu.Scroll to the bottom of this menu and you'll see a codes bar. Input your code here, being careful to make sure it’s in the right case and spelt correctly.Hit GO and your chosen code will be redeemed.Why Isn’t My Five Nights TD Code Working?If your Five Nights TD code isn’t working, it’s likely due to one of two scenarios. The first is that the code was inputted incorrectly. Make sure you have the code inputted exactly as it is in the Active Codes section above. You can even copy and paste codes directly from this article over to Roblox if you want to make sure you’re inputting them correctly.The other possibility is the code has expired or you’ve already used it. If you’ve used it, you’ll get a message inside the code bar reminding you that the inputted code has already been redeemed. If you’ve inputted it and it doesn’t recognise the code at all, it likely means it's no longer available to use and is expired.How to Get More Five Nights TD CodesIf you’re looking to scout out some Five Nights TD codes yourself, you should go ahead and join the game's Discord server. Once you're in, you'll find a channel titled "codes". The devs will post new codes in here as they add them, so keep an eye on it for the latest codes as they drop.What is Five Nights TD in Roblox?Five Nights TD is a tower defense game based on the Five Nights at Freddy's franchise. Like most tower defense games, the idea is to engage in strategic battles where iconic FNAF animatronics attack your home base. To defeat them, you need to place units to stop them in their tracks, tactically assigning familiar faces like Freddy, Chica, Bonnie, and more to quell the oncoming waves of robotic enemies.Along the way, you'll purchase new units, take on tougher difficulty levels, and face bosses, with the game pitting you against some of the franchise's most notorious antagonists.Callum Williams is an IGN freelancer covering features and guides. When he's away from his desk, you can usually find him obsessing over the lore of the latest obscure indie horror game or bashing his head against a boss in the newest soulslike. You can catch him over on Twitter at @CaIIumWilliams.