On-device speech-to-intent engine powered by deep learning
Switch branches/tags
Nothing to show
Clone or download



Made in Vancouver, Canada by Picovoice

Rhino is Picovoice's Speech-to-Intent engine. It translates speech commands into structured data representing user's intention. For example, given a speech command Can I have a small double-shot espresso with two sugars and no milk it will infer the intent and outputs the following structured data that can be used to take an action.

  "product": "espresso",
  "size": "small",
  "# shots": "double shot",
  "sugar": "two sugars",
  "milk": "no milk"

Rhino is

  • intuitive. It allows users to utter their intention in a natural and conversational fashion.
  • using deep neural networks trained in real-world situations.
  • compact and computationally-efficient making it suitable for IoT applications. It can run with as low as 100 KB of RAM.
  • cross-platform. Currently Android, iOS, Raspberry Pi, ARM Cortex-A, ARM Cortex-M, and a growing number of embedded platforms are supported.
  • customizable. It can be customized for any given domain (set of commands).

Table of Contents

Try It Out

Try out Rhino using its interactive web demo. You need a working microphone.


A good number of use-cases when building voice-enabled products revolves around understanding speech commands within a specific (limited) domain. For example, smart home alliances, mobile applications, etc. Rhino is a tight combination of speech-to-text and natural-language-understanding engines that are optimized to work for a specific domain. Rhino is quite lean and can run on small embedded processors with very limited RAM (as low as 100 KB) making it ideal for IoT applications. Furthermore, it can understand potentially unlimited number of commands within a specific domain. For example for coffee maker example above it can correctly recognize the following commands

  • can I have a latte?
  • make me a single-shot espresso.
  • I want a triple-shot americano with milk.
  • may I have a large cappuccino with cream?

Structure of Repository

Rhino is shipped as an ANSI C shared library. The binary files for supported platforms are located under lib and header files are at include. Bindings are available at binding to facilitate usage from higher-level languages/platforms. Demo applications are at demo. When possible, use one of the demo applications as a starting point for your own implementation. Finally, resources is a placeholder for data used by various applications within the repository.

Running Demo Applications

Running Python Demo Application

This demo application allows testing Rhino using computer's microphone. It opens an input audio stream, monitors it using Porcupine's library, and when the wake phrase is detected it will extract the intention within the follow-up command.

The following runs the Rhino engine to translate speech commands in the context of a coffee maker machine. Also, it initializes the Porcupine engine to detect the wake phrase Alfred. When the wake phrase is detected the Rhino starts processing the following speech command and prints out the inferred attributes and their values on the console.

python demo/python/rhino_demo.py --rhino_context_file_path=resources/contexts/coffee_maker.pv \

The following command runs the speech to intent engine within a smart light domain with wake phrase set to Rachel.

python demo/python/rhino_demo.py --rhino_context_file_path=resources/contexts/smart_light.pv \


Below are code snippets showcasing how Rhino can be integrated into different applications.


rhino.py provides a Python binding for Rhino library. Below is a quick demonstration of how to construct an instance of it.

library_path = ... # absolute path to Rhino's dynamic library
model_file_path = ... # available at lib/common/rhino_params.pv
context_file_path = ... # absolute path to Rhino's context file for the domain of interest
rhino = Rhino(

When initialized, valid sample rate can be obtained using rhino.sample_rate. Expected frame length (number of audio samples in an input array) is rhino.frame_length. The object can be used to monitor incoming audio as below.

def get_next_audio_frame():
    # implement the logic to get the next frame of audio

is_finalized = False

while not is_finalized:
    is_finalized = rhino.process(get_next_audio_frame())
    if is_finalized:
        if rhino.is_understood():
            for attribute in rhino.get_attributes():
                attribute_value = rhino.get_attribute_value(attribute)
            # logic to take action based on attributes and their values
            # logic to handle unsupported command

Finally, when done be sure to explicitly release the resources as the binding class does not rely on the garbage collector.



Rhinos is implemented in ANSI C and therefore can be directly linked to C applications. pv_rhino.h header file contains relevant information. An instance of Rhino object can be constructed as follows.

const char *model_file_path = ... // available at lib/common/rhino_params.pv
const char *context_file_path = ... // absolute path to Rhino's context file for the domain of interest
pv_rhino_object_t *handle;
const pv_status_t status = pv_rhino_init(model_file_path, context_file_path, &handle);
if (status != PV_STATUS_SUCCESS) {
    // error handling logic goes here

Now the handle can be used to monitor incoming audio stream. Rhino accepts single channel, 16-bit PCM audio. The sample rate can be retrieved using pv_sample_rate(). Finally, Rhino accepts input audio in consecutive chunks (aka frames) the length of each frame can be retrieved using pv_rhino_frame_length().

extern const int16_t *get_next_audio_frame(void);
while (true) {
    const int16_t *pcm = get_next_audio_frame();
    bool is_finalized;
    pv_status_t status = pv_rhino_process(handle, pcm, &is_finalized);
    if (status != PV_STATUS_SUCCESS) {
        // error handling logic goes here
    if (is_finalized) {
        bool is_understood;
        status = pv_rhino_is_understood(handle, &is_understood);
        if (status != PV_STATUS_SUCCESS) {
            // error handling logic goes here
        if (is_understood) {
            int num_attribtes;
            char **attributes;
            status = pv_rhino_get_attributes(handle, num_attributes, attributes);
            if (status != PV_STATUS_SUCCESS) {
                // error handling logic goes here
            for (int i = 0; i < num_attributes; i++) {
                char *attribute_value;
                status = pv_rhino_get_attribute_value(handle, attributes[i], &attribute_value)
                if (status != PV_STATUS_SUCCESS) {
                    // error handling logic goes here
                // logic to take an action based on attribute value
        else {
            // logic to handle out of context commands

When done be sure to release the resources acquired.



v1.0.0 November 2nd, 2018

  • Initial Release


Everything in this repository is licensed under Apache 2.0 including the contexts available under resources/contexts. Custom contexts are only provided with the purchase of the commercial license. In order to inquire about the commercial license send an email to contact@picovoice.ai with a brief description of your use case.