iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Yes—provided every link in the voice-control chain is local. A Home Assistant installation can receive speech, recognize it, decide what the command means, and operate compatible lights, switches, sensors, and appliances while the internet is disconnected. The home network must still be running, and each device must have a local integration or protocol.
An offline voice assistant is therefore a system, not a property of one speaker. A phone, dedicated satellite, or DIY microphone can be the room endpoint; Home Assistant normally runs on a separate, always-on host. Cloud-only devices, cloud speech recognition, remote-access services, online updates, and some third-party integrations stop working when their internet service is unavailable.
What “offline” means in a smart-speaker setup
There are two different networks to distinguish:
- Local network: your Wi-Fi or wired LAN, which lets the Home Assistant host, voice endpoint, and smart-home devices communicate inside the house.
- Internet: the outside connection used for vendor clouds, remote access, downloads, account services, and online speech or artificial-intelligence features.
An internet outage does not automatically disable a functioning local network. Home Assistant’s documentation says it can continue operating on your own hardware and lists Zigbee, Z-Wave, Matter, Thread, and ESPHome as examples of local device communication. The controller, network, and integrations still have to remain powered and reachable. See Home Assistant’s offline FAQ.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Conversely, a speaker that recognizes speech locally cannot make a cloud-only light local. If the light accepts commands only through the manufacturer’s server, it will be unreachable when that server or your internet connection is down.
#1 Best Overall
- Designed for Home Assistant Voice & Music Workflows: Preloaded with Home Assistant Voice Assistant and Music Assistant. Functions as both a voice input terminal and an audio playback endpoint.
- Dual Microphones for Voice Capture: Built with dual digital microphones for wake word or button-activated voice capture. Audio is streamed to the Home Assistant voice pipeline.
- Integrated 3W Speaker for Direct Playback: The built-in 3W/4Ω speaker supports TTS playback, Music Assistant streaming, and system audio without external speakers.
- Linux-Based Local Operation: Runs a lightweight Linux system on a quad-core ARM A53 CPU with 256MB RAM and 512MB flash for local audio processing.
- Development & Debugging Capabilities: Supports firmware flashing, and also provides access to live logs, on-device editing—suitable for routine development or issue diagnosis.
The local voice-control chain
A spoken command passes through several components. Offline operation requires a local path for each one:
- Endpoint: a microphone captures the user’s voice, and a speaker or media player can play the reply. The endpoint may be a phone, dedicated satellite, or DIY device.
- Wake-word and audio handling: the endpoint can listen continuously for a wake word or send audio after activation. Wake-word support varies by hardware and software.
- Speech-to-text: a local recognizer converts audio into text. This can be a constrained home-control engine or a more open-ended model.
- Intent handling: Home Assistant interprets the text as an action, such as turning on a light or setting a temperature.
- Local integration: Home Assistant sends the command to the device through a supported local protocol or integration.
- Text-to-speech: an optional local voice, such as Piper, turns the response into audio.
Home Assistant’s fully local voice guide describes this pipeline. A device entity also has to be exposed to Assist; an entity that is not exposed cannot be controlled through that voice setup. The Assist overview covers supported endpoints and voice features.
Which smart-home devices can work without internet?
Choose devices with a documented local integration or protocol, then verify compatibility for the exact model and Home Assistant integration. Common local communication choices named by Home Assistant include:
| Protocol or integration type | What it provides in an offline design |
|---|---|
| Zigbee | Local mesh communication through a compatible coordinator and Home Assistant integration. |
| Z-Wave | Local device control through a Z-Wave controller and integration. |
| Matter | Local control for compatible Matter devices when the required controller and network are available. |
| Thread | Local low-power networking; the Thread border-router and Home Assistant integration must remain available. |
| ESPHome | Local communication with ESPHome devices managed by Home Assistant. |
The protocol name alone is not a guarantee. A product may support a local radio while still requiring a vendor account for some functions, or its Home Assistant integration may expose only part of its capabilities. Check the integration’s current documentation and the exact model before buying. Devices that depend exclusively on a vendor cloud, cloud scene engine, or remote API cannot be controlled during an internet outage.
Local speech recognition: speed versus flexibility
Running speech recognition on your own hardware keeps audio and transcripts inside the local system, but the recognizer you choose affects command range, latency, language support, and hardware requirements.
| Option | Best fit | Published performance examples | Trade-off |
|---|---|---|---|
| Speech-to-Phrase | Fast, familiar home-control commands | Home Assistant reports transcription in under one second on Home Assistant Green or a Raspberry Pi 4. | Closed-ended: it supports a subset of Assist commands rather than unrestricted conversation. |
| Whisper | More open-ended speech and commands beyond a constrained grammar | Home Assistant reports around eight seconds on a Raspberry Pi 4 and under one second on an Intel NUC. | Needs substantially more compute for responsive results; performance depends on the actual host. |
| Piper | Local text-to-speech responses | Home Assistant reports 1.6 seconds of generated speech per second of processing on a Raspberry Pi using medium-quality models. | Voice quality and language coverage vary by language and model. |
These are hardware-specific figures, not universal latency promises. Speech-to-Phrase is usually the practical choice for lights, switches, scenes, and thermostats when speed matters. Whisper is the better fit when you need broader language understanding and have a more powerful local computer. Test the languages and accents you actually use; published support and quality can change with model updates.
Hardware roles: server, endpoint, and network
Home Assistant needs an always-on host. The official hardware page lists Home Assistant Green, Raspberry Pi systems, mini PCs or NUCs, virtual machines, and other supported computers. The hardware FAQ explains the host requirement. A host does not automatically give every room a microphone and speaker.
Recommended Free Tools
| Build path | Server role | Room endpoint | What to expect |
|---|---|---|---|
| Existing Home Assistant machine | Your current always-on computer, Raspberry Pi, mini PC, or virtual machine | Phone, dedicated satellite, or DIY endpoint | Lowest additional cost if suitable hardware is already running; you must configure the local voice pipeline. |
| Home Assistant Green host | Home Assistant Green | Separate phone or voice satellite | Plug-and-play host option, but Green is not itself a room microphone and speaker. |
| Voice Preview Edition | Separate Home Assistant host still required | The unit’s dual microphones and speaker | Dedicated endpoint with USB-C power, Wi-Fi, Bluetooth, audio output, physical microphone mute, ESPHome firmware, and open-source firmware for its ESP32 and XMOS chips. |
| ESPHome satellite | Home Assistant host | Supported ESP32-based microphone/speaker device | Flexible DIY route; audio processing is streamed to Home Assistant and resource-heavy configurations can exhaust microcontroller RAM or CPU. |
| Linux Voice Assistant | Home Assistant host | Compatible x64 or ARM64 Linux hardware | Experimental route described by Home Assistant; compatibility and setup require checking current guidance. |
The Voice Preview Edition documentation makes an important distinction: the product is a voice endpoint, not a standalone Home Assistant server. Its focused local processing handles common home-control phrases; full local processing requires a considerably more capable host for acceptable speed and accuracy. The official wake-word guide currently notes that Voice Preview Edition does not support custom wake words, so check current hardware documentation if a custom phrase is essential.
Rank #3
For ESPHome endpoints, the Voice Assistant component documentation says Voice Assistant support requires Home Assistant 2023.5 or later. It also warns that microphone, audio, and wake-word components consume meaningful resources and that heavy combinations can crash a microcontroller.
How to build a local voice system
- Keep the host available. Install Home Assistant on supported hardware and connect it to reliable power and your home LAN. If the host shuts down, voice control and automations that depend on it stop.
- Inventory device connectivity. For every light, switch, lock, climate device, or appliance, identify whether Home Assistant controls it locally or through a vendor cloud. Replace or isolate cloud-only devices if offline operation is a requirement.
- Add and verify local integrations. Pair the Zigbee, Z-Wave, Matter, Thread, or ESPHome devices using their supported Home Assistant integrations. Test control from the Home Assistant interface while the internet is still available.
- Expose only the entities you want to control. In the Assist or voice-assistant configuration, expose the relevant rooms, devices, and scenes. Unexposed entities will not respond to voice commands.
- Choose local speech services. Install a local speech-to-text engine and, if desired, Piper for local text-to-speech. Select Speech-to-Phrase for a narrow command set or Whisper when your host can handle open-ended recognition.
- Configure an endpoint. Use the companion app, a Voice Preview Edition, an ESPHome satellite, or another supported endpoint. Confirm that its microphone input reaches Home Assistant and that audio replies return to its speaker or media player.
- Test ordinary commands. Try commands for each exposed device, including room names, scenes, dimming, and temperature changes. Measure response time on the actual host rather than relying on figures from different hardware.
- Simulate an outage. Disconnect the internet while leaving the router, Wi-Fi, host, and smart-home hubs powered. Test speech recognition, device control, and replies. Then reconnect the internet and check whether any integrations require reauthentication.
What stops working when the internet is down?
- Cloud-only device control: commands cannot reach a device whose sole control path is the vendor’s server.
- Cloud speech or AI: any speech-to-text, intent, or text-to-speech service that sends audio or text to an online provider is unavailable.
- Remote access: Home Assistant Cloud and remote links through Apple Home, Google Home, or Amazon Alexa depend on internet connectivity.
- Updates and app-store access: downloading Home Assistant, firmware, language models, mobile apps, or integration updates requires internet access.
- Some language and advanced features: availability depends on the locally installed models and the capabilities of the integration; a feature offered online may not have an offline equivalent.
Offline does not mean “install once and never connect again.” Plan periodic internet access for initial installation, security fixes, firmware updates, model downloads, and recovery from a failed integration. During normal operation, however, local control can continue between those maintenance windows.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.ESPHome and dedicated satellites
ESPHome devices can stream microphone audio to Home Assistant for processing and receive audio through a speaker or media player. This architecture keeps the heavier recognition work on the host while allowing inexpensive room endpoints. It also makes resource planning important: audio buffers, wake-word detection, and other ESPHome components compete for microcontroller RAM and CPU. Start with the device’s supported configuration, add components gradually, and remove features if the endpoint becomes unstable.
Free tools Windows power users keep installed
One-click scans. No signup required.
A dedicated endpoint offers a more finished physical experience: built-in microphones, speaker, mute control, and a consistent enclosure. A phone is convenient for trying Assist without buying hardware. DIY Linux or ESPHome satellites can reduce cost or enable unusual room layouts, but require more wiring, firmware maintenance, and compatibility checks.
Rank #4
- Fitment For: Designed for Home Assistant Voice Preview Edition, ensuring a secure and stable wall mounting fit.
- Optimized Voice Interaction: Positions your device at a convenient height for better voice control experience.
- Stable Device Support: Provides firm support to keep the device securely positioned during daily use.
- Space Saving Wall Mount: Wall mount holder keeps your smart speaker off desks and counters for a cleaner setup.
- Multi Room Application: Ideal for kitchens, bedrooms, offices, and other smart home locations.
Where Rhasspy fits
Rhasspy describes itself as an open-source, fully offline collection of voice-assistant services that works with Home Assistant and other home-automation platforms. Its project documentation is aimed at technically capable users and describes splitting work between a more powerful base station and remote audio satellites.
That makes Rhasspy a legitimate offline architecture, especially when you want to assemble the speech services yourself. Its setup burden and project status should not be assumed to match a turnkey speaker: the referenced project pages do not establish a current release cadence or a support guarantee. Evaluate the current documentation, supported languages, and hardware compatibility before committing.
Privacy, reliability, and maintenance checks
- Map every network path. Confirm that microphone audio, transcripts, intent handling, and device commands stay on your LAN when local mode is selected.
- Use physical controls. A hardware microphone switch, such as the one on Voice Preview Edition, provides a clear way to cut microphone power when you do not want listening enabled.
- Keep local controls available. Wall switches, the Home Assistant dashboard, and manufacturer-independent controls provide recovery if speech recognition fails.
- Back up configuration. Store Home Assistant backups and document pairings, entity names, exposed devices, and wake-word settings before changing hardware or models.
- Plan for model and firmware updates. Updates may require internet access and can alter language support, resource use, or integration behavior. Test after updating.
- Separate outage types. An internet outage, a router failure, a Wi-Fi failure, and a Home Assistant host failure have different symptoms and fixes. A local voice system cannot operate if the LAN or host itself is down.
Which setup is right for you?
- Choose a phone endpoint if you already run Home Assistant and want to validate local Assist with minimal hardware.
- Choose a dedicated satellite if you want an always-available microphone and speaker in a room, with a physical mute control and less DIY work.
- Choose ESPHome or Linux DIY hardware if customization, multiple rooms, or experimentation matters more than plug-and-play setup.
- Choose Speech-to-Phrase when fast, predictable home-control commands are the priority and your phrases fit its supported subset.
- Choose Whisper when you need broader language understanding and can provide a host powerful enough to keep latency acceptable.
- Choose only locally integrated devices when internet-independent operation is non-negotiable; verify the exact model rather than relying on a protocol logo.
Final verdict
Open-source smart speakers can control devices without the internet, but only when the complete path is local: an operating Home Assistant host, a local speech pipeline, a working LAN, and devices controlled through local integrations. The speaker is merely the endpoint. Cloud-only hardware, cloud speech, remote-access services, and online maintenance remain outside that offline boundary.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
