Relay, a London-based startup founded by former Nothing employees, is launching a dedicated microphone called the Relay Q designed to capture whispered speech with high fidelity, addressing a gap in voice-to-text technology that existing software solutions haven't solved. The company debuted its macOS software today, with the hardware arriving in early 2027 at a price of $150, plus $15 per month for software after the first year.

Why Is a Dedicated Microphone Needed for Voice-to-Text?

The problem sounds simple but is surprisingly common in modern workplaces. Raymond Zhu, Relay's co-founder and former hardware product manager at Nothing, explained the frustration that inspired the company. "In the office, we never really could get the right volume for input," Zhu said. "If I whisper too quietly, the Mac wouldn't be able to pick it up; too loud and my coworker next to me can hear".

Raymond Zhu, Relay's co-founder and former hardware product manager at Nothing

While voice-to-text software has advanced rapidly thanks to large language models, most solutions rely on standard computer microphones or smartphone mics that struggle with quiet speech. Existing tools like Wispr Flow and Google's on-device dictation can intelligently strip filler words, clean up unscripted thoughts, and handle multilingual conversations, but they don't solve the acoustic challenge of capturing whispered input in shared spaces.

Relay's founders realized that the problem required hardware, not just software. Zhu noted that he would sometimes "pull out my earbud and go like this," holding it in front of his mouth to demonstrate the awkward workaround. "It calls for dedicated hardware," he explained.

Zhu

How Does Relay's Hardware and Software Work Together?

Relay's approach combines a dedicated microphone with intelligent software powered by Google's Gemini models. The cylindrical Relay Q microphone features a button on top that you hold down while speaking, similar to a walkie-talkie interface. The design allows you to hold it up to your mouth and whisper as quietly as you'd like, and Zhu says it should pick up what you say clearly.

The software works across text fields on your computer through a customizable hotkey, defaulting to the Fn key. When activated, Relay transcribes your speech and can perform contextual actions called "Skills," such as launching Slack to message a colleague or creating a calendar event in Google Calendar. The company offers preset skills for popular apps like WhatsApp and Gmail, and users can create custom skills to connect other applications.

In beta testing, Relay's transcription performed competently but not flawlessly. It generally understood when users misspoke and restated sentences, cutting out the irrelevant parts. It could also insert emoji when requested, though this feature worked inconsistently during testing. The software can be customized to match individual communication styles; for example, users can instruct it to avoid adding periods in Slack messages or to translate messages into a colleague's native language.

What Are the Key Features and Customization Options?

• Contextual Awareness: Relay can distinguish between messages and separate actions, such as asking it to send a message and provide a Google Maps link in the same breath, though this feature can be inconsistent in beta.
• Personalization by Contact: The software detects who you're communicating with and allows you to set different tones; messages to your boss can sound professional while messages to friends remain casual.
• Language and Style Customization: Users can add notes in the Relay app to establish patterns, such as avoiding periods in certain contexts or translating messages for international colleagues.
• Local Data Storage: Relay stores data locally on the user's device, where users can manually edit or delete it, though data sent for processing passes through Relay's servers in transit to Google's servers.

The company is also creating a separate MagSafe grip that allows the Relay Q to attach to the back of an iPhone or Qi2-enabled device, enabling voice-to-text on mobile devices when the hardware launches in the first quarter of 2027.

What Privacy and Security Trade-Offs Should Users Know About?

Using Relay's automation features requires granting the app sweeping system permissions. Beyond microphone access, users can allow Relay to record a snapshot of their screen whenever it's activated, which enables the software to execute Skills like finding a contact in Slack or creating calendar events. The company states it does not retain data after processing, but granting AI models visibility into your operating system introduces privacy and security considerations similar to Microsoft's Recall feature.

Relay is exclusive to macOS at launch, though the company plans to expand to Windows, iOS, and Android. The company chose to start with desktop because Zhu believes conversations are most complex in that environment. "Chats with friends and family are simple on phones, but at work in front of a computer, you're sharing ideas, brainstorming, and using varied tone and formality," he explained.

How Does Relay Compare to Other Voice-Input Hardware?

Relay is not the first company to recognize the potential of dedicated hardware for voice input. Last year, a startup called Sandbar unveiled the Stream Ring, and Pebble later showed off its own take, the Index 01. Both are rings designed for users to whisper their thoughts into them, and both are starting to ship to initial buyers. Zhu noted that Relay considered a ring format but ultimately decided against it, especially given the crowded smart ring market.

The Relay Q will come with a charging dock for desktop storage when not in use. The company's vision centers on a simple question: "How long can you go with your workday without touching your keyboard?" This philosophy drives both the hardware design and the software's automation capabilities, positioning Relay as a tool for users who want to minimize keyboard interaction throughout their day.