Voice AI News — Aug 10–17, 2026
" was in my thought but not in the count? Ah, "multimodal model." -> "m-u-l-t-i-m-o-d-a-l- -m-o-d-e-l-." Let's use Python-like len() on: "Stay updat
Whissle AI Combines Transcription, Emotion and Intent Understanding
Whissle's META-1 speech technology combines multilingual transcription with features such as emotion and intent understanding, providing richer information from spoken conversations than transcription alone.
Thinking Machines Lab Releases Open-Weight Inkling Model
Thinking Machines Lab introduced Inkling, a 975-billion-parameter open-weight multimodal model capable of working with text, images, audio and video.
Apple Negotiates Publisher Deals to Bring Current News to Siri AI
Apple is reportedly discussing compensation agreements with publishers to give its next-generation Siri access to timely news and information.
NVIDIA Opens Early Access to Nemotron 3 VoiceChat
NVIDIA introduced Nemotron 3 VoiceChat, a 12-billion-parameter full-duplex speech-to-speech model designed for natural, interruptible, low-latency voice-agent conversations.
PNB Housing Finance Tests Voice AI for Collections and Loan Conversion
PNB Housing Finance has been experimenting with conversational and voice AI across customer engagement and collections as part of its broader digital transformation.
KAIST Develops Neuromorphic Chip for Speech Processing
KAIST researchers developed a neuromorphic computing approach that uses semiconductor noise to process signals including speech and motion more efficiently.
Zymbly Builds Voice-First Copilot for Aircraft Maintenance
Y Combinator-backed Zymbly is developing a voice-first AI copilot that helps aircraft technicians troubleshoot problems, work with maintenance manuals and turn spoken notes into documentation.
Microsoft Retires Its Mico Companion From Copilot Voice
Microsoft is removing the animated Mico AI companion from Copilot Voice less than a year after the character's introduction.
AI Voice Acting Used to Localize Assault on Dark Athena
Mechanics VoiceOver released an AI-assisted localization of The Chronicles of Riddick: Assault on Dark Athena after traditional voice-acting funding efforts fell short.
HP Partners With Sarvam AI for Multilingual Voice Experiences
HP and Sarvam AI signed an agreement to bring multilingual Indian language and voice AI capabilities directly to HP devices.
Kansas Officials Warn Residents About AI Voice-Cloning Scams
Officials in Mount Hope, Kansas warned residents about scams in which AI-generated voices imitate family members to make convincing emergency money requests.
Modi's Independence Day Speech Translated With Bhashini AI
India's Bhashini platform was used to translate Prime Minister Narendra Modi's Independence Day speech into 22 Indian languages.
Wippi Raises $1.2M for Screen-Free Voice AI Products for Children
Bengaluru-based Wippi raised $1.2 million in seed funding to expand its screen-free, voice-based learning products for children.
Sarvam Makes Its Voice Agents Platform Publicly Available
Sarvam opened its Voice Agents platform to developers and businesses building enterprise-scale conversational experiences for Indian users.
FreJun Launches Teler for Programmable Voice AI Telephony
FreJun's Teler provides programmable calling infrastructure, real-time audio streaming and APIs for developers building AI phone agents.
Hippocratic AI Launches Agentic Orchestrators for Healthcare Voice AI
Hippocratic AI introduced orchestration technology designed to coordinate teams of conversational AI agents around healthcare workflows and patient outcomes.
Cozyla Debuts Voice-Controlled Smart Family System
Cozyla introduced family organization devices powered by a voice agent that helps households manage schedules, routines and shared tasks.
Regal Integrates Autonomous Voice Agents With Five9
Regal partnered with Five9 to integrate autonomous AI voice agents into enterprise contact-center workflows and customer interactions.
Cancer Survivor Reclaims Her Voice Using AI Cloning
Cancer survivor Alice Harty used AI voice-cloning technology to recreate a version of her own voice after losing the ability to speak naturally.
Vocaloop Launches Multimodal Layer for Voice Agents
Vocaloop introduced technology that allows voice AI agents to use visual interfaces alongside speech, addressing tasks that are difficult to complete through voice alone.
Google Meet Adds AI Notes for In-Person Meetings
Google Meet is expanding automated note-taking so speech recognition can capture discussions happening in physical meeting rooms as well as online calls.
Resemble AI Expands Its Focus on Deepfake Detection
Resemble AI is increasingly focusing its technology on detecting manipulated audio, video and images as synthetic media becomes harder to distinguish from authentic content.
ElevenLabs Partners With Ominimo for AI Voice Customer Support
Digital insurance company Ominimo is working with ElevenLabs to bring realistic conversational voice assistants into its customer-service operations.
AI Voice Analysis Shows Potential for Detecting Early ALS Signs
Research suggests that AI analysis of speech characteristics could help identify vocal changes associated with amyotrophic lateral sclerosis.
IISc Releases SraVaani for Indian Speech Recognition
IISc researchers introduced SraVaani 1.0, an open multilingual ASR model designed to broaden speech-recognition support across 65 Indian languages and dialects.
sensiBel and Aizip Partner to Improve Speech Recognition in Noise
sensiBel and Aizip combined optical MEMS microphone technology with edge AI to improve speech capture and recognition in noisy environments.
Deepgram Introduces Flux TTS for Real-Time Voice Agents
Deepgram's Flux TTS is a streaming-first text-to-speech family built specifically for conversational voice agents, including cross-turn context and real-time delivery.
SAFC Adopts Conversational AI for Collections
South Asialink Finance Corp. is deploying AI Rudder's conversational technology to strengthen customer engagement and automate parts of its collections operations.
Telefónica Adds Generative AI to Business Voice Services
Telefónica is integrating AI transcription, summarization and intelligent assistants into business voice services in Spain.
SoundHound Reports Revenue Growth While Losses Continue
SoundHound reported strong year-over-year revenue growth in the second quarter while the Voice AI company continued to operate at a loss.
AI Voice Clone Raises Evidence Questions in Utah Murder Case
Prosecutors in a decades-old Utah murder case have explored using AI to recreate victim Joyce Yost's voice from existing recordings and sworn testimony, raising novel questions about synthetic evidence in court.
Consumers Show Growing Acceptance of AI Customer-Service Agents
More sophisticated conversational agents are beginning to replace basic scripted chatbots as businesses attempt to make automated service interactions feel more useful and natural.
InnoCaption CEO Calls for Accessibility to Be Built Into AI Infrastructure
InnoCaption CEO Paul Lee argues that accessibility features such as high-quality transcription should be treated as core AI infrastructure rather than optional add-ons.
Fonio.ai
Reaches €10M ARR With AI Voice Agents - Vienna-based reported reaching the €10 million annual recurring revenue milestone while expanding AI phone agents for European small and medium-sized businesses.
Rootle Expands Voice AI Platform for India's BFSI Sector
Rootle expanded its enterprise Voice AI offering to support TRAI-compliant customer communications for banks and other financial institutions in India.
Singapore Hospital Scales Speech-to-Text for Nurses
National University Hospital is using AI-enabled transcription to reduce nursing documentation work and make clinical workflows more efficient.
iFLYTEK Demonstrates Real-Time Multilingual Translation Screen
A transparent display demonstrated real-time speech recognition and language translation at an AI experience center in China.
Sarvam AI Launches Indic DiarBench Across 22 Indian Languages
Sarvam AI and AI4Bharat introduced an open benchmark designed to evaluate speech recognition and speaker attribution across 22 Indian languages.
Twilio Reports Sharp Growth in Voice AI Adoption
Twilio says self-service voice usage is accelerating as more companies move conversational AI deployments from pilots into production.
Oink Ink Heads Launches Professional Voice-Cloning Business
The broadcast production company is expanding into professional AI voice cloning as synthetic voice technology moves further into commercial media production.
Apple Expands Siri AI Voice Customization
Apple's iOS 27 beta reportedly adds controls that let users adjust aspects of Siri's speaking pace and expressivity.
Alibaba Launches CosyVoice Studio Speech Platform
Alibaba introduced CosyVoice Studio as a full-stack speech AI environment bringing together voice-generation tools and its broader Qwen-Audio ecosystem.
Cornerstone Insurance Adds Voicebot to Its CiCi Assistant
Cornerstone Insurance introduced voice interaction for its CiCi virtual assistant to expand automated customer support beyond text.
North Korean Hackers Experiment With AI Transcription
Cybersecurity researchers report that the Kimsuky threat group has experimented with AI tools to transcribe stolen calls and meeting recordings.
Salesforce Launches Agentforce Voice in Japan
Salesforce introduced Agentforce Voice in Japan, using local-language technology to automate more contact-center conversations and address staffing pressure.
KT Builds Next-Generation AI Contact Center for NH Nonghyup Bank
KT completed a major AI contact-center project designed to modernize consultation systems at NH Nonghyup Bank.
ElevenLabs Expands Emotion-Preserving AI Dubbing
ElevenLabs' latest dubbing technology is designed to translate spoken content across dozens of languages while retaining more of the original speaker's delivery and emotion.
USC Researchers Find Audio AI Still Struggles to Read Emotional Cues
USC researchers found that multimodal AI systems can perform substantially differently when interpreting emotional information from audio compared with written text.
New Orleans Uses AI to Assist With Some 911 Calls
New Orleans is using AI in a limited role to assist with some emergency calls, while human dispatchers remain the primary responders; reports that AI had replaced 911 dispatchers were inaccurate.
AI Companies Are Building Digital Replicas of Deceased People
New services are using photos, recordings and other personal media to create interactive digital replicas that can reproduce the voices of people who have died.
Harley-Davidson Dealership Deploys AI Voice Assistant After Hours
Bulldog Harley-Davidson is using an AI assistant to answer incoming calls outside normal business hours and respond to common customer requests.
AI Voice Journaling Apps Turn Spoken Thoughts Into Interactive Reflection
Voice-first journaling products such as Journee are using conversational AI to help users capture and work through thoughts without traditional typing.
TTSWP Brings AI Text-to-Speech to WordPress Sites
The TTSWP plugin uses ElevenLabs voices to convert website articles into spoken audio across more than 70 languages.
CliniSense AI Automates Clinical Skill Assessment
Researchers developed a pipeline that converts clinical audio into competency-aligned assessments of resident performance.
Uzbek Speech Recognition Improved Through Model Adaptation
Researchers showed that transfer learning and language-model fusion can improve Wav2Vec2 XLS-R automatic speech recognition for Uzbek.
Persona-ASR Targets Overlapping and Code-Switched Speech
Researchers proposed a two-stage Kazakh-English target-speaker ASR system designed to isolate the intended speaker and reduce errors during overlapping speech.
Role-Based Voice Agents Counter Unwanted Telemarketing
A study proposes narrowly constrained voice agents that can interact with unwanted telemarketing calls while following explicit task boundaries and escalation rules.
Virtual News Anchor Framework Integrates Speech Synthesis
Researchers designed an automated news-anchor system combining news acquisition, text-to-speech generation and lip-synchronized video.
Multimodal Screen-Share Voice Assistant Targets Sub-700ms Responses
Researchers developed a browser-based assistant combining screen context with real-time WebRTC voice interaction for low-latency multimodal assistance.
Smart Meeting Twin Rehearses Voice Interviews
A research prototype uses speech recognition and multiple AI personas to conduct voice-first practice interviews alongside meeting-management features.
AS-Split Conformer-Mamba Improves Long-Utterance Speech Recognition
The proposed architecture separates local and global speech modeling to improve recognition and alignment across long spoken sequences.
Multimodal Embeddings Retrieve Caregiver-Child Behaviors Without Transcripts
Researchers evaluated multimodal embedding models for finding behavioral moments directly inside caregiver-child audio and video streams.
Ocean Announcer Uses Indonesian Text-to-Speech for Marine Forecasts
The STMKG-OcA system automatically retrieves marine weather forecasts and converts them into Indonesian speech for coastal communities.
AI Folk Music Platform Evaluates Vocal Performance
Researchers developed a system that uses speech technology to assess pitch, timbre and ornamentation in traditional folk singing.
Speech Emotion Recognition Classifies Indonesian Stress Levels
A hybrid CNN-LSTM and IndoBERT system analyzes Indonesian voice notes to classify psychological stress levels.
Secure Voice Authentication System Uses Biometric Speech Recognition
Researchers developed a banking authentication prototype that identifies users through distinctive vocal characteristics.
Visual Speech Recognition App Targets Deaf Users
An Indonesian web application uses visual lip information rather than audio to recognize spoken language and address challenges such as homophones.
Vocal Variation Degrades Automatic Speaker Recognition
A replication study found that changes in how the same person speaks can reduce the reliability of modern speaker-recognition encoders.
Folk Tale Framework Synchronizes Synthetic Speech and Lip Motion
Researchers created a multimodal framework combining emotion-driven speech synthesis with synchronized facial and lip animation for traditional stories.
Conformer Architecture Powers Automated Clinical Documentation
A clinical documentation framework uses a 12-layer Conformer speech architecture to transform spoken clinical information into structured SOAP-style records.
Arabic Speech Resource Targets ASR Errors in Sacred Texts
Researchers released a unified Arabic speech dataset and benchmark containing hadith audio to improve recognition of religious material.
Wireless Arduino Gloves Translate Sign Language Into Speech
Researchers developed gloves that translate hand gestures into synthesized speech while also converting spoken responses into displayed text.
Medical Voice Translator Addresses Spanish Dialect Differences
A prototype clinical translation system routes spoken medical communication according to community-specific Spanish vocabulary and dialect variations.
Nani AGI Architecture Integrates Continuous Voice Interaction
The proposed architecture combines language-model reasoning with speech recognition, language identification and speech synthesis for continuous interaction.
CRNN-V2 Improves Speech Emotion Recognition
Researchers developed a model combining convolutional networks, bidirectional recurrent units and attention mechanisms to improve emotion classification from speech.
AI Interview Assistant Uses Speech and Emotion Recognition
A research prototype analyzes spoken interview answers and emotional characteristics to provide candidates with personalized performance feedback.
Bimodal Framework Combines Voice and Text for Emotion Recognition
Researchers combined acoustic information with predicted textual sentiment to improve recognition of emotions in speech.
Android App Translates Speech Into Indonesian Fingerspelling
A feasibility study presents a mobile application that recognizes spoken words and maps them into Indonesian sign-language fingerspelling.