BHASHINI and IISc build Sourashtra text-to-speech models
India's BHASHINI platform and IISc Bengaluru developed text-to-speech models for the Sourashtra community in Tamil Nadu, expanding speech synthesis to a low-resource language.
BHASHINI
Latest in AI Voice rss
Fish Audio releases Drama 3 directable AI voice model
Drama 3 is Fish Audio's voice model for directable AI voice acting, with natural-language control over emotion, pacing, whispers, laughter and pauses across 83 languages. On web, mobile, API, with open weights.
Launch · ▲ 83 on Product Hunt
Fish AudioOpenOpen weightsAPIApp
Comparison pits Grok Voice 3.0 against Qwen-Audio-3.0-TTS
OrcaRouter published a comparison of Grok Voice 3.0 and Qwen-Audio-3.0-TTS, contrasting how the speech systems hear versus render audio.
Ghana judiciary to buy speech-to-text for local languages
Ghana's Chief Justice says the judiciary will procure speech-to-text machines to transcribe local languages into English by year-end.
Sarvam AI releases Bulbul V4 Flash text-to-speech
Why it matters: Adds fast, India-focused TTS options to the voice-agent stack.
Bulbul V4 Flash is Sarvam AI's new fast text-to-speech model for Indian languages, bringing voice-agent campaigns and analytics into coding tools.
Sarvam AIClosedAPI
More: RuntimeWire
Stellon Labs adds 1.7B speech model and paid APIs to KittenML
KittenML now offers a larger speech model, hosted text-to-speech and transcription APIs and emotion-aware transcription, alongside the founders' existing 25 MB CPU speech model.
Stellon LabsOpenAPIOpen weights
SoundHound AI launches Kia voice AI
SoundHound AI rolled out voice AI for Kia vehicles; analysts set a $12.75 price target on the stock.
SoundHound AIClosedApp
Built In explains ElevenLabs AI voice platform
An explainer on ElevenLabs, the AI voice platform for text-to-speech, voice cloning and dubbing. It describes what the product does and how creators use it; no new release or facts.
ElevenLabsClosed
China's DeepZang AI model targets better Tibetan translation
DeepZang, released in March, converts speech and writing between Tibetan, Mandarin Chinese and English, and will undergo a series of linguistic upgrades, Sixth Tone reports.
DeepZangOpen
Vobiz AI builds telecom infrastructure for voice AI
Vobiz AI offers telecom infrastructure and connectivity for voice AI agents, handling telephony, SIP and call routing so developers can deploy real-time voice applications.
Vobiz AI
CareCycle launches free Medicare plan comparison voice tool
Voice AI startup careCycle rolled out a free plan comparison tool for Medicare beneficiaries, using voice AI to help them evaluate and choose coverage options.
careCycleFree
Hungarian language dictation support finally available
A user reports that Hungarian language dictation support is finally available, referring to an app's speech-to-text feature rather than a formal product announcement.
Superwhisper 3.0 launches bot-free offline meeting recorder
Superwhisper, a dictation tool with LLM reformatting for macOS, Windows and Linux, reached version 3.0 with a meeting recorder that joins calls without a bot and can run offline.
Version 3.0.0 · macOS / Windows / Linux
SuperwhisperClosedAppPaid
More: RuntimeWire
OpenAI rolls out GPT-Live-1.1 ChatGPT voice update
Why it matters: Voice is a primary ChatGPT interface; new multilingual voices widen its reach.
OpenAI released GPT-Live-1.1 for ChatGPT voice, adding six new voices including Korean and Hindi-focused options, more natural chat and time limits.
OpenAIClosedAppFreePaid
Geek Vibes Nation: real-time voice AI fits geek entertainment
Geek Vibes Nation argues real-time voice AI belongs in geek and entertainment culture, describing dubbing and translation use cases for films, games and fandom content.
Disabled peer uses AI to recreate her former voice in Lords speech
A disabled peer used AI voice-recreation technology to deliver a speech to the House of Lords in the voice she used to have, per the Disability News Service.
Identity
Ubuntu 26.10 debuts Myna, a local AI voice agent
Neowin reports Ubuntu 26.10 ships Myna, a local AI voice agent named after a talking bird, letting users run voice interactions on-device without cloud services.
CanonicalOpenFreeCode
MWire Labs shows Garo speech-to-text AI model at Google event
Syllad reports Shillong-based MWire Labs showcased what it says is the first public Garo speech-to-text AI model at a Google event, aiming to support the low-resource language.
MWire LabsOpen
Kathmandu Post asks how well AI fakes Nepali voices
The Kathmandu Post speaks to experts about how convincingly AI can clone Nepali voices, and what that means for fraud and media trust in Nepal.
Fraud & riskIdentity
Grok Voice 3.0 tops Voice AI Index but lags on support calls
H2S Media reports Grok Voice 3.0 leads the Voice AI Index yet still underperforms on real customer-support calls, where latency and task completion matter more.
xAIClosedAPI
Maya 2 is a free open-weight voice model for 22 Indian languages
Why it matters: It broadens open speech coverage for a large, under-served language market.
Maya 2 is an open-weight voice AI model covering 22 Indian languages, released free for speech tasks such as text-to-speech and voice interaction.
MayaOpenOpen weights
Edifier launches We-Bass open-ear earbuds with DeepSeek AI
Edifier launched We-Bass open-ear earbuds with integrated DeepSeek AI features, bringing AI voice assistance to an open-ear audio product.
EdifierOn sale
Afterparty launches agents that draft meeting follow-up actions
Afterparty is a Product Hunt tool that turns meeting transcripts into drafted emails, messages, invites and CRM notes across connected apps, with human approval before sending.
Launch · ▲ 83 on Product Hunt
AfterpartyClosedAppPaid
Grok Voice resolves 70% of Starlink customer calls
Why it matters: Shows real-time voice agents handling commercial customer service at scale.
Grok Voice reportedly resolves 70% of Starlink's customer calls with no human intervention, as businesses weigh how to verify outcomes and measure cost per case.
xAIClosedAPI
More: Business Insider, AOL.com
NatWest and Nvidia adapt speech AI to UK regional accents
NatWest worked with Nvidia to make its AI spending tool recognize UK regional accents better, improving speech recognition for customers.
NatWestClosed
More: Finextra · All 2 sources →
Atticus launches push-to-talk voice control for Mac
Atticus is a Mac app where you hold a key, speak, and the computer performs the requested action, using voice to control desktop tasks.
AtticusClosedApp
Druid AI adds voice, interoperability and governance to agent platform
Druid AI expanded its enterprise agentic AI platform with new voice capabilities plus agent interoperability and governance features for enterprise deployments.
Druid AIClosed
Inworld launches realtime inference for consumer AI apps
Why it matters: Realtime low-latency inference is key for consumer voice and character AI products.
Inworld launched realtime inference aimed at consumer AI applications, targeting low-latency voice and interactive experiences for developers.
InworldClosedAPI
More: 01net
xAI launches Grok Voice 3.0 speech-to-speech model for voice agents
Why it matters: A frontier lab's new real-time voice stack raises the bar for conversational agents.
Grok Voice 3.0 is xAI's new speech-to-speech model for building voice agents, handling Starlink and Tesla calls. API pricing is unchanged and existing users are auto-migrated.
xAIClosedAPIOfficial
Square launches Kiosk with conversational AI order help
Why it matters: Puts voice AI ordering hardware directly into merchants' hands as agentic commerce grows.
Square launched the Square Kiosk, a pill-shaped touchscreen ordering device with an agentic conversational voice layer that understands a seller's menu and listens to spoken orders.
SquareClosedOn saleAgentic commerce
Samsung expands Galaxy AI features to more languages
Samsung is rolling out its Galaxy AI tools, including Transcript, Call and Note Assist, to more users and additional languages.
SamsungClosedApp
SoundHound AI launches in-car voice AI in India
SoundHound AI launched its first in-car AI product in India, with analysts debating what the move means for the company's rich valuation.
SoundHound AIClosed
BreezeBlue-TTS-2 claimed best local TTS but non-commercial
A Reddit user says BreezeBlue-TTS-2 is the best locally runnable text-to-speech model right now, beating Moss TTS 1.5, but its license is non-commercial. Moss TTS 1.5 remains best free-to-use.
BreezeBlueOpenOpen weights
Tether backs AIsa, brings on-device speech recognition to glasses and robots
Why it matters: Combines stablecoin payments for AI agents with on-device speech hardware, targeting agentic commerce.
Tether invested in AIsa and announced on-device speech recognition for smart glasses and humanoid robots. The push aims to enable stablecoin AI-agent payments and private edge voice AI.
TetherClosedAgentic commercePayments
Maya-2 open-weight Indic voice model claims top performance
Why it matters: A competitive open-weight Indic voice model broadens multilingual speech AI options.
An X post announces Maya-2, billed as the world's best Indic voice AI model with fully open weights, claiming it outperforms ElevenLabs v3, Cartesia Sonic 3.5 and Bulbul v3 across Indian languages; free to builders soon.
♥ 12 on X
Maya-2OpenOpen weights
Stripe highlights Gradium and CreativeFitting voice and video AI payments
Stripe customer stories describe Gradium using Stripe to bill for global voice AI services and CreativeFitting scaling AI-generated short dramas internationally with Stripe payments. Both are case studies, not new products.
StripeOfficialPayments
Apple researchers detect code-switching via language-selective dimensions
Apple ML researchers propose finding code-switched utterances in large speech collections using language-selective dimensions in speech foundation models, helping add them to training data.
Apple
Fish Audio releases Drama 3 open-weight TTS model
Why it matters: Adds a large open-weight MoE contender to the fast-moving text-to-speech race.
Drama 3 is a 35B-parameter mixture-of-experts text-to-speech model with about 3B active parameters. Fish Audio released the weights, and users on Reddit and MarkTechPost call it a new leading open TTS option.
Fish AudioOpenOpen weights
More: MarkTechPost, Reddit
UK lawmaker delivers first parliament speech with AI voice
A British lawmaker became the first person to use an AI-generated voice to deliver a speech in the UK parliament, with software recreating her former voice during a health debate to read a speech she wrote herself.
Grok for iPhone 1.4.53 released
xAI updates its Grok iOS app with improvements to Chat, Voice and Imagine.
Version 1.4.53 · iPhone / iPad
xAIClosedAppFreePaid
UK Parliament hears first AI-voiced speech via ElevenLabs
Baroness Campbell addressed the House of Lords using her own AI-recreated voice built with ElevenLabs, the first AI-voiced speech in UK Parliament history.
ElevenLabsClosedOfficial
Developers turn EmbeddingGemma 2 into a speech-to-text model
A project repurposes Google's EmbeddingGemma 2 embedding model into a speech-to-text system, with the write-up and demo shared on Hacker News.
OpenOpen weightsCode
Gladia to join OVH Groupe for European sovereign voice AI
Voice AI company Gladia is set to join OVH Groupe, aiming to accelerate Europe's sovereign voice AI capabilities with European-hosted speech technology.
Gladia
Deepgram targets voice AI on air-gapped laptops
Why it matters: Local voice AI on air-gapped machines widens deployment for privacy-sensitive industries.
Deepgram is expanding its Flux voice model to 10 languages and targeting air-gapped laptops, letting voice AI run locally without cloud connectivity for sensitive environments.
DeepgramClosedAPI
Google tests new Gemini Live mobile UI with waveform animation
Google is testing a redesigned Gemini Live mobile interface with a new waveform animation and a dedicated voice output control.
GoogleClosed
Tesla brings Grok Bot voice tasks to Cybercab and Model Y vehicles
Tesla is bringing the Grok Bot to Cybercab and robotaxi fleet vehicles, enabling voice-controlled tasks during rides, and expanding Grok Bot voice tasks to Model Y.
TeslaClosed
TII releases Falcon-ASR, a 1.6B Arabic speech recognition model
TII releases Falcon-ASR, a 1.6B-parameter Arabic speech recognition model with an Emirati focus, reporting 20.92% average WER on six test sets. Weights and API are not yet available.
TII
Rivian rolls out new AI voice assistant to its vehicles
Rivian is rolling out a new AI voice assistant in its vehicles, with InsideEVs detailing what it can do.
RivianClosed
Atticus, a hold-to-talk macOS voice assistant running on Claude, shared on Hacker News
Atticus is a Mac voice assistant: hold Control-Option-Space, speak or type a task, and it runs it, sending simple tasks to cheaper handling. Built by a 9th-grade developer and shown as Show HN.
App
Open WebUI v0.12.0 adds realtime voice calls
Open WebUI is a self-hosted, user-friendly chat interface for Ollama and OpenAI-compatible APIs. v0.12.0 adds realtime voice calls using an OpenAI Realtime voice model, switchable by administrators in audio settings.
Release v0.12.0 · open-webui/open-webui
Open WebUIOpenCodeFree
EMA Lightning releases 8.6M-parameter Turkish text-to-speech model
EMA Lightning is a tiny Turkish TTS system: a 5.6M-parameter DiT acoustic model plus a 3M-parameter vocoder, about 34 MB total, Apache 2.0. It claims 0.92% WER on Freya-TR-Eval and 3.86 ms first audio.
Model · text-to-speech · apache-2.0 · ♥ 330 · 15.7k downloads
canberkkkkkkOpenOpen weights
SoundHound AI stock falls as Nvidia-backed voice firm slides
Benzinga reports SoundHound AI shares are in freefall, examining why the Nvidia-backed voice AI company's stock has dropped sharply.
SoundHound AI
Anthropic may be developing its own voice models, TestingCatalog reports
TestingCatalog reports that Anthropic might be developing its own voice models. No launch or official confirmation has been reported.
AnthropicClosed
Gmail Live may gain seven new Gemini voices, reports Android Authority
Google's Gmail Live feature may soon add seven new voices powered by Gemini, according to Android Authority. The change is not yet rolled out.
PocketWebTools launches Mac app for local AI chat, transcription and video
PocketWebTools for Mac bundles local AI chat, transcription and video tools in one app that runs on your Mac. Launched on Product Hunt.
Automation Anywhere acquires Boost.ai for voice AI
Why it matters: Consolidation among enterprise automation vendors pursuing autonomous customer-facing agents.
Automation Anywhere agreed to acquire conversational voice AI company Boost.ai from Nordic Capital to expand customer-facing voice agents and its 'autonomous enterprise' push.
Automation Anywhere
Willow Voice picks CoreWeave for AI model training support
Willow Care, maker of AI dictation app Willow Voice, chose CoreWeave for model training, citing forward-deployed support rather than raw compute. The app combines speech recognition with dictation features.
CoreWeave
OpenAI rolls out Dots to all mobile users, letting them voice-command Codex
OpenAI's Dots persistent agent is now fully rolled out on mobile, letting users issue voice commands to Codex for task handling. The Register reports open-source imitators have followed.
More: eu.36kr.com