🧔♂️ A friendly human may check it before it goes live. More news here
Camb.AI, Broadcom to run speech models on device chips
Camb.AI and Broadcom have announced a collaboration to run CAMB.AI’s text-to-speech AI models on Broadcom’s system-on-chip platforms.
Camb.AI, a US-based AI localization startup, has developed a feature that converts written text into speech in multiple languages, now available on devices powered by Broadcom chips.
The text-to-speech function operates directly on Broadcom’s integrated neural processing unit, allowing the technology to work without cloud connectivity and potentially improving privacy and reducing latency.
The companies said this approach could benefit applications such as customer service, e-learning, and accessibility features for visually impaired users.
The partners are also exploring whether CAMB.AI’s real-time language translation model can be run on Broadcom’s on-device hardware to support translation in over 150 languages.
Broadcom is a major supplier of chips for home entertainment and connectivity devices.
Camb.AI has previously worked with organizations such as Comcast NBCUniversal and IMAX, and has raised US$18.5 million to date.
🔗 Source: Camb.ai
🧠 Food for thought
Implications, context, and why it matters.
Broadcom SoC platforms with integrated neural processing units (NPU) bring on-device AI to home entertainment
- Broadcom supplies set-top box (TV receiver) chips plus broadband system-on-chip (SoC) parts for home entertainment and connectivity gear 12. In 2015, it presented a TV dongle to cable providers that delivered full cable with digital video recorder (DVR) features 3.
- Running CAMB.AI’s text-to-speech on the on-chip NPU keeps audio private and cuts delay, since no audio goes to cloud servers.
- Wi‑Fi issues for behind‑TV dongles 3 can arise from blocked signals. That can hamper updates or hybrid cloud and device processing even if the AI runs locally.
Device makers can stand out with multilingual voice features without recurring cloud costs
- Smart TV and streaming device makers could add accessibility features like screen reading or voice navigation in many languages without ongoing cloud Application Programming Interface (API) fees. The partners are exploring on-device real-time translation in over 150 languages on Broadcom hardware.
- Independent software vendors (third-party app developers) that build for operator platforms (TV platforms run by cable and telecom providers) could use on-device translation to localize customer service without hardware refresh cycles. Providers such as Comcast have evaluated Broadcom’s device concepts 3.
- Broadband providers that adopt Broadcom processors with up to 5 Gbps downstream (download) speeds 3 could bundle AI voice services as premium features. That can open new revenue without infrastructure overhauls.
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.