Search “Speech”

Sunmi SDC2026 Developer Conference concludes: launches Hi Sunmi agent and Windows commercial terminals

Sunmi Technology held the SDC 2026 Developer Conference in Shanghai, launching the business-specific AI agent "Hi Sunmi," which supports voice interaction, business self-diagnosis, customer flow analysis, and intelligent marketing. On the hardware side, Sunmi partnered with Intel to launch Windows-architecture commercial terminals such as the FLEX 32 WIN and T3 PRO WIN, enabling parallel Android and Windows ecosystems. The Sunmi Magic AI agent platform has served nearly 40,000 merchants, and Sunmi's medical insurance facial-recognition terminals have been deployed in over 150,000 medical institutions, with more than 350,000 devices deployed.

When smartwatches and other devices listen in: what is legally permitted for recording and transcription

A heise legal guide points out that recording and processing speech in Germany without the consent of all participants may simultaneously violate data protection law and Section 201 of the Criminal Code, the offense of "breach of the confidentiality of speech." Lawyer Jessica Flint advises obtaining permission from everyone present before recording, and preferably also explaining the intended use of the subsequent transcription. The article also discusses whether Apple Watch's Audio Intelligence feature can be used legally in Germany.

Alibaba Cloud Model Studio Token Plan personal edition starts at 39 yuan/month: Lite plan includes 11,500 Credits

The lowest price for the Alibaba Cloud Model Studio Token Plan personal edition is 39 yuan/month, corresponding to the Lite plan (original price 60 yuan/month, limited-time discounted price 39 yuan/month). It includes 11,500 Credits per month and supports 1-2 concurrent Agents. This plan covers models for text generation, image generation, video generation, speech recognition, and real-time voice conversation. The quota is calculated by subscription month and does not roll over. After it is used up, you can purchase a usage pack for 100 yuan/month that includes 20,000 Credits (up to 5 can be held at the same time). Currently, only the China North 2 (Beijing) region is supported.

OpenAI Opens Decisions API to All, Competing with TypeSafe's Jev for the Decision Model Market

OpenAI opened the Decisions API to all customers on Thursday; the product was announced at the September 29 Developer Day and is based on a reconfigured ChatGPT Luna to speed up inference and reduce costs. Nikunj Handa, OpenAI's head of API product, said latency has been cut from 1 to 3 seconds to 100, 120, and 180 milliseconds; the public version adds voice and vision modalities and deploys inference resources by region to reduce latency. In early trials at Box, Datadog, Ramp, and Speak, Speak cut the action interval from 1.4 seconds to 0.13 seconds; on the AIM System One Index and the Artificial Analysis Decision Index, OpenAI's decision model ranks eighth and fourteenth, respectively, trailing Liquid's D1 Ultra and Base Lab's Elephant-1.

Nirva raises $33 million and launches a pendant device that only listens to its wearer

Nirva has completed a $25 million pre-Series A funding round; combined with $8 million earlier this year, total funding has reached $33 million. Its first device is a pendant weighing about 8 grams that can be worn as a bracelet, with up to 3 days of battery life. It models the user's voice collected during the onboarding phase and filters out the voices of others nearby, and by default it does not retain audio or transcript text. The device is priced at $249, with pre-orders opening in November and shipments to early users before the end of the year.

Red and Black List of Four Personal Agents: Grok Bot, Muse, Claude Duet, and ChatGPT Dots Tested by Division of Labor

Peter Yang updated his tested red and black list of four personal agents: Grok Bot excels at real-time X data research but its email feature is unusable; Muse is free and stands out in offline business phone capabilities but is slightly weaker in raw intelligence; Claude Duet has the strongest proactivity and base-model intelligence but its research preview lacks browser takeover; ChatGPT Dots can inherit historical assets but still has bugs and slow voice connections. He connects no more than 4 agents to his personal email at the same time.

Synaptics Launches Astra SR200 Series Edge AI MCUs with Integrated 256 MB Custom DRAM

Synaptics has released the Astra SR200 series of edge AI MCUs, combining heterogeneous AI-native computing with integrated high-bandwidth custom DRAM scalable to hundreds of megabytes, delivering up to 400 GOPS of AI acceleration and 256 MB of custom DRAM, and integrating Arm Cortex-M85 and Cortex-M52 cores. The series includes three pin-compatible models—SR210 (200 GOPS), SR230 (300 GOPS), and SR250 (400 GOPS)—targeting wearables, connected cameras, and industrial vision and voice systems.

Anthropic launches Duet, an always-on personal agent

Anthropic has released Duet, an always-on personal agent that runs as an app separate from Claude on iOS, Android, and the web. Dan Shipper said that after testing it for two weeks, the share of his time spent using Claude rose to about 50%; the app includes channels divided by work and life domains, to-do lists, and document and slide editing features. He also noted that Duet cannot yet direct Claude sessions on a computer, that its email drafts are not as good as dot's, and that its voice feature only supports transcription and cannot hold a conversation.

Peter Yang reviews Grok Bot, Muse, Claude Duet, and ChatGPT Dots

Peter Yang has updated his review of four personal agents: Grok Bot is free and integrates with X, with excellent voice mode and Pages features, but its email function is basically unusable; Muse is free and offers the best consumer-grade experience, with standout performance on business calls; Claude Duet is the most proactive out of the box and has the strongest model capabilities, but the research preview lacks features such as driving a browser; ChatGPT Dots can drive cloud and local threads within ChatGPT, but still has bugs and slow voice connections.

HeyGen Video and Voice Launch on fal Platform

fal announced that HeyGen's HeyGen Video and HeyGen Voice are now available on fal. Both models have built-in audio: they support text, image, and reference video to generate videos with dialogue, ambient sound, and sound effects, with clip lengths of 5 to 15 seconds and up to 2K; Voice supports 39 languages, instant voice cloning, and offers a faster Turbo tier.

Anduril investor MVP Ventures plans to raise $100M opportunity fund

MVP Ventures is in talks to raise $100 million to increase its stakes in its best-performing companies. The five-year-old firm has raised two venture funds and made small early-stage investments in Palmer Luckey's Anduril, Erebor, and ModRetro, as well as AI training startup Turing, voice AI company Wispr Flow, and rocket launch startup Stoke Space.

What Is Alibaba Cloud Model Studio Token Plan: Credits Billing and Subscription Pricing Explained

Alibaba Cloud Model Studio has launched the Token Plan large model subscription service, which uses Credits as a unified unit of measurement, supports models for text, image, and video generation as well as speech recognition, and can be used in tools such as Claude Code, Cursor, and Qwen Code. Limited-time pricing for the Personal edition ranges from Lite at 39 yuan/month (11,500 Credits) to Pro at 499 yuan/month (180,000 Credits), while the Team edition starts at 150 yuan/seat/month (25,000 Credits) for a standard seat. Credits are dynamically deducted based on model type, Token usage, thinking mode, and tool calls; once a plan's quota is used up, it does not switch to pay-as-you-go billing.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

Gradium open-sources the Phonon on-device TTS model suite, with all models under 50m parameters

Gradium open-sources the Phonon on-device speech synthesis model suite, with all models under 50m parameters, focused on multilingual naturalness and accuracy, and claims to offer the broadest inference support to date. Phonon-31m is released under the Apache 2.0 license. Matt Turck reposted this news.

English translation is being prepared.