Voice Recognition Module V3: specs and pinout
A speaker-dependent voice recognition module: you train it with short spoken commands, load up to seven of them, and it reports the matched record ID over a 5 V TTL serial port. Recognition runs on the module, and trained commands are kept in its flash memory.
Specs
| Spec | Value | Source |
|---|---|---|
| Supply voltage | 4.5 V to 5.5 V (regulated) | source Product specifications, p.3; 'Before connecting power', p.2 |
| Supply current | Less than 40 mA (the manual gives no measurement conditions) | source Product specifications, p.3 |
| Logic level | 5 V TTL UART and GPIO | source Product specifications, p.3 |
| Serial default | 9600 bit/s, 8 data bits, no parity, 1 stop bit | source Connect a USB-TTL adapter, p.6 |
| Stored voice records | Up to 80 (record IDs 0 to 79) | source Product specifications, p.3 |
| Simultaneously active records | Up to 7 | source Product specifications, p.3 |
| Maximum command length | 1500 ms (a short word or phrase); each record is trained with two recordings | source Product specifications, p.3; training steps, p.6 |
| Microphone input | 3.5 mm mono microphone connector and a microphone pin interface | source Product specifications, p.3 |
| Board size | 31 mm x 50 mm | source Product specifications, p.3 |
Power
| Value | Source | |
|---|---|---|
| Supply | 4.5 V to 5.5 V, regulated | source Product specifications, p.3 |
| Current | Less than 40 mA | source Product specifications, p.3 |
Pinout
Labels are from the maker's user manual (R4, 15 Oct 2026). The manual says its photographs are for orientation and are not a dimensioned footprint or a connector pin-number specification, so use the labels printed on your board.
| Pin | Function | Voltage | Notes | Source |
|---|---|---|---|---|
| VCC / GND | Module supply and common ground | 4.5 V to 5.5 V | source Interfaces and labels, p.3 | |
| TXD | UART output from the module (cross to the host receiver) | 5 V TTL | source Interfaces and labels, p.3 | |
| RXD | UART input to the module (cross to the host transmitter) | 5 V TTL | source Interfaces and labels, p.3 | |
| IN0-IN2 | External group-selection inputs | 5 V TTL | source Interfaces and labels, p.3 | |
| OUT0-OUT6 | Recognition-related digital outputs | 5 V TTL | The manual says to confirm the slot-to-output mapping before integration. | source Interfaces and labels, p.3 |
Build with it
Project ideas for the Voice Recognition Module V3. Each link opens Soldr with the prompt filled in; Soldr assigns the pins, checks power and shows how to wire every part.
- Voice-controlled indicator lights: The Uno is a 5 V board, matching the module's 5 V supply and 5 V TTL serial. The module takes under 40 mA. Train each word twice and keep it under 1.5 s. Add a series resistor to every LED.
- Voice-triggered sound player: Both modules use TTL serial at 9600 bit/s by default, so the Uno needs two serial ports (a software serial port for one of them). The record ID in the recognition reply selects which track to play.
- Voice command monitor on an LCD: The module reports the record ID (and the signature label you trained) when it matches, which the sketch prints. Good for checking how reliably your own words are recognized before using them to drive something.
- A voice-controlled desk lamp switch that turns a low-voltage 5 V LED lamp on and off with two spoken commands using an Arduino Uno, a voice recognition module V3 and a relay module: Train 'on' and 'off' as two records. Keep the load to the low-voltage LED circuit on the relay's switch side; wiring a mains lamp must be done by a qualified person and is not part of this idea. Learning project, not a security or safety device.
Common mistakes
- Using a 3.3 V board directly: the manual states the serial interface and GPIO are 5 V TTL and the supply is 4.5 V to 5.5 V, so a 3.3 V host needs level handling or a 5 V board such as an Uno (source: link, p.2-3).
- Trusting a USB-TTL adapter's 5 V pin to mean its signal pins are 5 V tolerant: the manual says it does not, and the adapter must be compatible with 5 V TTL (source: link, p.2 and p.6).
- Expecting recognition right after training: a trained record only takes part in recognition after it is loaded into one of the seven slots, and the loaded set must be loaded again after a power cycle unless auto load is configured (source: link, p.4).
Not covered by the official source
- Operating temperature range is not given in the maker's manual.
- Pin numbers and the exact connector layout are not given; use the labels printed on the board.
- The 99 % recognition figure in the manual has no stated test conditions and is not used on this page.
Sources
- link (ELECHOUSE Voice Recognition Module V3 user manual, R4, 15 Oct 2026; maker host)