SoundHound Acquires LivePerson

May 26, 2026

🔊 Soundcheck

  • SoundHound Acquires LivePerson

  • AudioHijack Threatens Voice AI Security

  • Palabra.ai Hits $1M ARR Fast

  • iOS 27’s AI Voice Control Sparks Siri Reboot

Read time: 4 minutes

🔥 Hot Mic

Big moves, deep dives, and standout stories.

SoundHound AI is acquiring LivePerson to create a unified voice‑and‑digital AI platform targeting $500M in revenue. This move brings together SoundHound’s advanced voice and agentic AI with LivePerson’s digital messaging platform, creating an end‑to‑end omnichannel conversational AI solution. The combined business taps LivePerson’s one billion messages per month capacity and SoundHound’s voice intelligence, aiming to serve enterprises across industries and streamline customer interactions across voice, text, and web. Financially, the deal values LivePerson at $43M equity with $250M enterprise value, and positions SoundHound for aggressive growth toward $350M–$400M in 2027 revenue and a $500M cross‑sell potential. The acquisition is expected to close in H2 2026 pending approvals, adding scale, a broader customer base, and deep integration between voice and digital channels.

Key Points:

  • Unifies voice agentic AI and digital messaging into one platform.

  • Valuation: $43M equity, ~$250M enterprise value.

  • Targets $350M–$400M 2027 revenue with $500M cross‑sell upside.

  • Expected to close in second half of 2026, pending approvals.

Takeaway: By merging voice and digital conversational AI under one roof, SoundHound is building a powerful omnichannel platform that could leapfrog legacy providers—and potentially unlock half‑billion‑dollar revenue opportunities.

Researchers have revealed a stealthy form of attack called AudioHijack, where imperceptible audio alterations can coerce voice AI systems to perform unintended actions like web searches, file downloads, or sending emails, even though humans don’t notice anything wrong.

This threat isn’t limited to academic experiments. AudioHijack targets powerful generative audio-language models that can listen, summarize, act, and use tools. It works across numerous open models and even impacts production systems from major providers.

The risk is urgent for any voice interface that can act—such as meeting assistants or smart agents. Traditional defense methods like prompting or intent-checking barely mitigate the threat. Developers must rethink how they trust audio input and build safety measures that prevent hidden commands from becoming harmful actions.

Key Points:

  • AudioHijack uses nearly imperceptible audio tweaks to inject commands.

  • Attack tested on 13 open models and commercial systems like Azure, Mistral.

  • Success rates ranged from 79 percent to 96 percent across scenarios.

  • Simple defenses reduced risks only modestly, so new safeguards are essential.

Takeaway: As voice AIs gain the ability to act on audio inputs, developers must stop treating audio as inherently safe—hidden prompts are real attacks and deserve serious, multi-layered defenses.

Palabra.ai, based in London and backed by Seven Seven Six, surged from approximately $60,000 in annual run rate in October 2025 to $1 million by April 2026. Their real-time speech translation works instantly in over 60 languages and preserves the speaker’s original voice, offering seamless communication across meetings, webinars, and live broadcasts. The platform packages speech recognition, translation, and text-to-speech into a single, low-latency API, with SDKs in Python, JavaScript, and Java. It's GDPR-compliant, ISO 27001-certified, and never stores customer audio, ensuring privacy and security. Palabra’s clients include big names like DHL, UNICEF, Hyundai, BCG, Deloitte, Fujitsu, DocuSign, eToro, and Agora. Use cases range from live meeting translation on Zoom, Teams, and Meet to QR-based in-person listen-on-phone setups—at around one-tenth the cost of a human interpreter.

Key Points:

  • Annual run rate jumped 17× in six months—from ~$60K to ~$1M.

  • Real-time translation in 60+ languages and 1,000+ language pairs.

  • Voice cloning works from just six seconds of speaker audio.

  • Platform GDPR‑compliant, ISO 27001 certified, no audio stored.

Takeaway: Palabra.ai demonstrates that real-time, voice‑preserving AI translation is not just feasible—it’s rapidly becoming indispensable for global teams, offering high-speed, secure, and affordable multilingual communication.

iOS 27 introduces natural language voice control, hinting at a richer, context-aware Siri overhaul. Apple’s latest accessibility enhancement in iOS 27 brings a major leap with AI-powered Voice Control. It lets users operate their iPhone or iPad using natural descriptions—like saying “tap the purple folder”—instead of needing exact labels. This new capability ushers in the deeper Siri transformation Apple has been teasing: an assistant that understands context, sees what's on your screen, and takes meaningful actions across apps. It’s a powerful preview of what’s to come with Siri’s long-promised AI upgrade.

Key Points:

  • Voice Control now supports natural language descriptions of on‑screen elements

  • You can say things like “tap the guide about best restaurants”

  • Feature powered by Apple Intelligence and eases accessibility

  • Signals upcoming Siri upgrade with on-screen awareness and context

Takeaway: The new AI-driven Voice Control in iOS 27 primes Siri for a major shift—moving from limited commands to a smarter, context‑aware assistant that truly understands your screen and intent.

🎙️ Mic Drop

What else is making noise in voice AI.

Demonstrates how hidden, inaudible sounds in online audio streams can manipulate and compromise voice AI chatbot security. (futurism.com)

Researchers warn background ‘hidden’ audio in media can prompt unauthorized actions in voice AI agents—posing real-world attack risks. (cybernews.com)

Compares top voice generator tools, with detailed strengths for business, dubbing, narration, and estimated voice AI market growth projections. (letsdatascience.com)

Curates leading voice agents like Retell, CloudTalk, Synthflow, with a focus on efficiency and integrations for SaaS operations. (technology.org)

Synthflow AI now powers 5M+ calls monthly, illustrating enterprise migration to natural conversation and automation over keypad IVR. (tech.eu)

A practical guide to reducing end-to-end latency in real-time voice AI applications—crucial for call quality and user experience. (technology.org)

New global competition focuses on turn-taking modeling, encouraging innovation in conversational voice AI interruptions and flow. (aithority.com)

Industry perspective on how AI is impacting voice routing, telecom margins, and operational models for voice service providers. (telecomreseller.com)

Examines the rise of AI-powered voice recorders that offer transcription, summarization, and productivity-enhancing integrations for business users. (cybernews.com)

Coros emphasizes integrating AI-powered voice for sports wearables, highlighting potential for hands-free, context-aware health tracking. (thenews.com.pk)

Landmark lawsuit on AI voice use, copyright, and deepfake legislation—raises compliance questions for voice model builders using celebrity likenesses. (letsdatascience.com)