Voice AI Weekly # 2 - Third Week of July 2026
We have selected five noteworthy overseas voice AI news stories from the past week. This week, the 'places' where voice AI is deployed expanded, reaching on-device, networks, and even hospital sites.
1. Deepgram Runs Nova-3 on Snapdragon-Powered PCs
Deepgram has enabled its speech recognition model, Nova-3, to run directly on Qualcomm Snapdragon X series PCs. This method processes real-time speech recognition on the device's Hexagon NPU without going through the cloud. Developers and manufacturers can now integrate voice features without worrying about response speed, privacy, or network stability. The trend of the voice AI industry, which used to be cloud-first, moving towards the edge has become distinct. This seems to be a signal that the market is opening up for companies that have prepared full-pipeline NPU porting and on-device licensing.
Source: Telecom Reseller
2. Rime Secures $24M Series A - "Speech-to-Speech for Enterprise"
Voice AI startup Rime has raised $24 million in a Series A round led by M13. Rime processes nearly 100 million phone calls per month, with Mayo Clinic and Dialpad among its clients. The investment will be used to expand its proprietary conversational datasets and research staff. Notably, the stages of its growth have been healthcare and finance. The market where pronunciation accuracy of specialized terminology and regulatory compliance constitute competitiveness seems to be growing.
Source: Morningstar (Business Wire)
3. LG Uplus and Ericsson Jointly Develop Network-Based Voice AI
On July 14, LG Uplus signed a partnership for voice AI and AI-native network collaboration at Ericsson's headquarters in Stockholm. By combining the commercial data of its AI call agent ixi-O with Ericsson's network technology, they aim to create a structure that provides voice AI at the telecommunications network level, rather than being confined to apps or OSs. The goal is a 'Zero UI' environment where intent is conveyed through voice alone. When the call itself becomes an AI touchpoint, what is deciphered from the call audio can determine service quality. The utility of voice signal analysis technology seems to have expanded to the level of telecommunications infrastructure.
Source: Mobile World Live
4. Research Reveals Voice Changes Predict COPD and Asthma Exacerbations Early
Research has been published showing that worsening asthma and COPD symptoms can be detected several days in advance solely through changes in voice recorded on smartphones. The research app, TACTICAS, tracks signal changes in short voice recordings left daily by patients to capture signs of deterioration. This means that the voice signal itself, rather than the spoken content, reveals the health status. Clinical research is accumulating evidence of acoustic intelligence, demonstrating what information is contained in the signal layer prior to text transcription.
Source: Medical Xpress
5. UK NHS Deploys Ambient Voice Technology to 70,000 Clinicians
NHS England Midlands has completed procurement of Ambient Voice Technology (AVT) for 1,239 GP practices, 15 trusts, and over 70,000 clinicians. The goal is to automatically transcribe conversations during consultations so that doctors can focus on patients instead of paperwork. This is the first large-scale regional deployment in the NHS. In highly regulated environments like healthcare, voice AI is beginning to transition from pilot projects to standard infrastructure.
Source: HTN
If you have any further questions about today's news, or if you are considering introducing voice technology into your business, please feel free to email us at contact@holamago.com.
Wishing you an energetic and productive week ahead.