OpenAI’s API Drives Audio AI Phase-shift

TECH IN THE NEWS

$100T and Rising - The IMF forecasts global public debt to exceed $100T by the end of year and warned major economies’ plans to stabilise borrowing “fall far short.”

"Click to Cancel" - The FTC announced a new rule that will require companies to make it as easy to cancel subscriptions to their services as it is to sign up.

ECB Cuts - The European Central Bank has cut interest rates by a quarter point to 3.25%, amid signs of slowing growth and weakening inflation in the Eurozone.

More Chip Blocks - The US may cap chip exports to countries, particularly in the Middle East, and lawmakers seek to block Huawei from buying chipmaking gear.

Trump-onomics - In a Bloomberg interview, Trump defended sweeping tariff plans, criticized Fed Chair Powell, said president should have “input” on rate decisions, says he won't let Nippon Steel buy US Steel, and called TikTok a "security threat.”

Windows-GPT - OpenAI is launching a dedicated ChatGPT app on Windows. It’s currently in preview to select users but should roll out more broadly later this year. 

COMPANIES TO WATCH

AI Going Nuclear - Amazon joins Google and Microsoft in tapping nuclear energy sources to power its AI, with three new energy deals including leading a $500M round for SMR startup, X-energy.

Lightmatter - The California startup building photonic tech in datacenter networking chips raised a $400M Series D at $4.4B led by T. Rowe Price joined by existing backers Fidelity and GV.

Boston Dynamics - The MA. robotics pioneer partnered with Toyota Research Institute to bring TRI’s large behavior model tech to BD’s Atlas electric humanoid.

Blackstone - The NYC asset manager emerged as a major global data center funder with a new $8.2B deal in Spain. It beat Q3 profit estimates as its AUM hit a record.

What do Nvidia and Amazon have in common?

Well, other than trillion-dollar market caps. Both Nvidia and Amazon chose to collaborate with Miso. Miso’s the leader in AI-powered kitchen robots. That’s why Nvidia offered Miso its premier AI vision tech, and Amazon handpicked Miso to partner and use its RoboMaker simulation environment. Now, Miso launched their first commercial AI-powered robot, and it sold out in seven days. On the back of that success, they’re focused on scaling to 170+ US fast food brands in need.

Invest in Miso by 10/18.

THE HOTTEST THING IN TECH
The Audio AI Phase Shift

Yesterday, OpenAI extended its audio transcription-to-synthesis (TTS) layer to developers via its API. This means apps built on the GPT4o model can now accept audio inputs, and offer audio responses. The explosion in audio content and interfaces over the last few years has been rapid, but ways of dealing with all that media behind the scenes hasn’t quite caught up. AI is changing that fast with a number of new startups, as well as the big tech players rushing to corner the Voice AI market. AI executive Gary Grossman sees this AI phase shift as highlighting a trend of AI evolving to replicate human-like behaviors, “a development that will soon reshape how businesses work, and how workers engage with technology.”

The Tech Base 

The transcription (input processing) to synthesis (output generation) layer is essential for audio AI systems, enabling seamless verbal communication between humans and machines. It also unlocks AI sentiment analysis, information extraction, and conversation summarization. TTS applications are vast and expanding. The main ones going forward will likely be AI Agents, Digital Co-pilots and Voice Assistants (e.g., Alexa 2.0 and Siri 2.0), which can understand and respond to user commands. It also helps with content creation (audiobooks, virtual narration, and personalized audio advertisements) and even accessibility tools for people with disabilities.

Startup Solutions 

Audio AI Startup Gladia raised $16 million in Series A funding this week and released its “Real-Time” API solution. It bridges the gap between accuracy and speed in real-time transcription, ensuring low latency (under 300 ms) without compromising accuracy and ability to provide real-time insights beyond transcription.

Another startup in the space is Assembly AI. Powered by a $28M raise earlier this year, Assembly is aiming at becoming the go-to solution for analyzing speech, offering ultra-simple API access for transcribing, summarizing and otherwise figuring out what’s going on in thousands of audio streams at a time.

Big Tech Enters the Game

OpenAI's new approach to audio AI with GPT4o and the API audio update changes the game in several ways for the TTS layer and makes life difficult for startups like Gladia and Assembly. Its OpenAI Codex and Whisper models (for transcription) and GPT-based generative tools are already widely used for audio tasks. And, now developers can incorporate them into their own apps.

These innovations are geared toward making the transcription>>synthesis step faster, more efficient, and importantly, more natural-sounding and context-aware. It achieves this through end-to-end models, diverse and massive datasets resulting in faster, adaptable systems with more human understanding and generation. 

Google is also innovating heavily. Its Duplex uses Automatic Speech Recognition (ASR) to convert spoken requests into text and Text-to-Speech (TTS) synthesis to generate a natural-sounding response based on the query, matching conversational context. DeepMind’s WaveNet synthesis layer generates raw waveforms that approximate human speech, giving much more realistic output.

Market Size and Share

The global AI audio transcription-to-synthesis market is projected to hit ~$5.21B by 2027 with a CAGR of over 21.5% from 2020-2027. Longer-term, some estimates suggest this market will surpass $27B by 2030, driven by increasing adoption in industries like media, education, and enterprise communication. Key Investors in the space include some of the smartest money around, Andreessen Horowitz, Kleiner Perkins, Tiger Global, Insight Partners and Sequoia. 

How Big A Slice Will The Big Kids Take?

Obviously any AI application involving massive datasets and intensive compute loads is going to skew heavily in the favor of tech giants like OpenAI, Microsoft and Google. The question is, what slice of the pie will be left for the smaller startups? 

Given OpenAI’s extensive partnerships and dominant presence, it is likely capturing a significant portion of the market, particularly with its Microsoft Copilot partnership, through which it powers solutions like Azure’s transcription and AI video tools. It likely holds a significant share of the market, though this figure may increase with its recent API update and as it expands product offerings and enterprise system integrations.

TRENDING TOOLS AND BUZZY TECH
Tools

Apple - New 7th gen iPad mini at $499 will support AI + AI Smart Glasses by ‘27.

Amazon - New AI-powered advertiser creative suite + a Kindle color e-reader.

Adobe - ‘Project Super Sonic’ AI tool text-to-SFX gen + object ID, voice imitation.

Emteq - ‘Mind Reading’ Smart Glasses monitor emotions, eating habits + more.

Mistral - New "edge device” optimized models for phones, laptops connect to LLMs.

Tech

Lyten - US startup to invest over $1B in Nevada lithium-sulfur battery factory.

MIT - Solar-powered desalination system requires no extra batteries, power.

Lacoste - Taps AI for anti-counterfeit tech to uncover fakes at 99.7% accuracy.

NTHU - World’s smallest quantum computer solves problems with just 1 photon.

ORNL - US develops lightest crack-free alloy that can withstand 2,400°F heat.

380x Faster Than 5G - Scientists set new wireless transmission speed record.

FOR INVESTORS
Open Deals

Request an intro to founders - invest@thetech.buzz

1. 🔥⏱️ Robotics for Agricultural Efficiency - Pittsburgh - Built & deployed robotic hardware for the largest U.S. tree nurseries, reducing costs and increasing efficiency by 50x. Hardware and AI data layer. Beginning to scale globally. Post-revenue, at capacity. (Seed+)
   

2. 🔥⏱️ Quant Trading Software Company - New Jersey - Large-scale automated day-trading algorithm designed for volatile U.S. equity markets. Profitable through extensive backtesting and live simulation. Seeking equity investment for software company.

3. Cricket for the Modern World - India - Physical and virtual ecosystem fostering cricket experiences for enthusiasts and upcoming pros. Data-driven, locally proven, globally scaling. (Seed)

🔥 — hot deal!
⏱️ — leaving soon.

FOR EMPLOYERS
Job Candidates

Request an intro to candidates - jobs@thetech.buzz.

1. Struck Capital Venture Associate, ex-West2East (Russell Wilson) Director of Special Projects, ex-Phoenix Holdings Investment Associate. Harvard BA. Vietnamese pro basketball player.

2. Lockheed Martin Software Engineer, ex-Rutgers Junior Full Stack Developer. Secret clearance. Expertise in C++, PHP, SQL. Rutgers BA in Computer Science.

🚨TECH BUZZ ANNOUNCEMENT:

We’ve made some changes — Our Closed Deals, Companies to Watch, and Funding Buzz sections will now arrive every Monday AM wrapping up the week’s top deals.

Please get in touch with feedback on this change and any other suggestions as to what you would like us to feature more of.