Skip to content

User Guide

To get started with WebRTC streams, all that's needed is to import the WebRTC component from this package and implement its stream event.

This page will show how to do so with simple code examples. For complete implementations of common tasks, see the cookbook.

Audio Streaming

Reply on Pause

Typically, you want to run an AI model that generates audio when the user has stopped speaking. This can be done by wrapping a python generator with the ReplyOnPause class and passing it to the stream event of the WebRTC component.

ReplyonPause
import gradio as gr
from gradio_webrtc import WebRTC, ReplyOnPause

def response(audio: tuple[int, np.ndarray]): # (1)
    """This function must yield audio frames"""
    ...
    for numpy_array in generated_audio:
        yield (sampling_rate, numpy_array, "mono") # (2)


with gr.Blocks() as demo:
    gr.HTML(
    """
    <h1 style='text-align: center'>
    Chat (Powered by WebRTC ⚡️)
    </h1>
    """
    )
    with gr.Column():
        with gr.Group():
            audio = WebRTC(
                mode="send-receive", # (3)
                modality="audio",
            )
        audio.stream(fn=ReplyOnPause(response),
                    inputs=[audio], outputs=[audio], # (4)
                    time_limit=60) # (5)

demo.launch()

  1. The python generator will receive the entire audio up until the user stopped. It will be a tuple of the form (sampling_rate, numpy array of audio). The array will have a shape of (1, num_samples). You can also pass in additional input components.

  2. The generator must yield audio chunks as a tuple of (sampling_rate, numpy audio array). Each numpy audio array must have a shape of (1, num_samples).

  3. The mode and modality arguments must be set to "send-receive" and "audio".

  4. The WebRTC component must be the first input and output component.

  5. Set a time_limit to control how long a conversation will last. If the concurrency_count is 1 (default), only one conversation will be handled at a time.

  1. The python generator will receive the entire audio up until the user stopped. It will be a tuple of the form (sampling_rate, numpy array of audio). The array will have a shape of (1, num_samples). You can also pass in additional input components.

  2. The generator must yield audio chunks as a tuple of (sampling_rate, numpy audio arrays). Each numpy audio array must have a shape of (1, num_samples).

  3. The mode and modality arguments must be set to "send-receive" and "audio".

  4. The WebRTC component must be the first input and output component.

  5. Set a time_limit to control how long a conversation will last. If the concurrency_count is 1 (default), only one conversation will be handled at a time.

Reply On Stopwords

You can configure your AI model to run whenever a set of "stop words" are detected, like "Hey Siri" or "computer", with the ReplyOnStopWords class.

The API is similar to ReplyOnPause with the addition of a stop_words parameter.

ReplyonPause
import gradio as gr
from gradio_webrtc import WebRTC, ReplyOnPause

def response(audio: tuple[int, np.ndarray]):
    """This function must yield audio frames"""
    ...
    for numpy_array in generated_audio:
        yield (sampling_rate, numpy_array, "mono")


with gr.Blocks() as demo:
    gr.HTML(
    """
    <h1 style='text-align: center'>
    Chat (Powered by WebRTC ⚡️)
    </h1>
    """
    )
    with gr.Column():
        with gr.Group():
            audio = WebRTC(
                mode="send",
                modality="audio",
            )
    webrtc.stream(ReplyOnStopWords(generate,
                            input_sample_rate=16000,
                            stop_words=["computer"]), # (1)
                  inputs=[webrtc, history, code],
                  outputs=[webrtc], time_limit=90,
                  concurrency_limit=10)

demo.launch()
  1. The stop_words can be single words or pairs of words. Be sure to include common misspellings of your word for more robust detection, e.g. "llama", "lamma". In my experience, it's best to use two very distinct words like "ok computer" or "hello iris".
  1. The stop_words can be single words or pairs of words. Be sure to include common misspellings of your word for more robust detection, e.g. "llama", "lamma". In my experience, it's best to use two very distinct words like "ok computer" or "hello iris".

Stream Handler

ReplyOnPause is an implementation of a StreamHandler. The StreamHandler is a low-level abstraction that gives you arbitrary control over how the input audio stream and output audio stream are created. The following example echos back the user audio.

Stream Handler
import gradio as gr
from gradio_webrtc import WebRTC, StreamHandler
from queue import Queue

class EchoHandler(StreamHandler):
    def __init__(self) -> None:
        super().__init__()
        self.queue = Queue()

    def receive(self, frame: tuple[int, np.ndarray]) -> None: # (1)
        self.queue.put(frame)

    def emit(self) -> None: # (2)
        return self.queue.get()

    def copy(self) -> StreamHandler:
        return EchoHandler()


with gr.Blocks() as demo:
    with gr.Column():
        with gr.Group():
            audio = WebRTC(
                mode="send-receive",
                modality="audio",
            )

        audio.stream(fn=EchoHandler(),
                     inputs=[audio], outputs=[audio],
                     time_limit=15)

demo.launch()
  1. The StreamHandler class implements three methods: receive, emit and copy. The receive method is called when a new frame is received from the client, and the emit method returns the next frame to send to the client. The copy method is called at the beginning of the stream to ensure each user has a unique stream handler.
  2. The emit method SHOULD NOT block. If a frame is not ready to be sent, the method should return None.
  1. The StreamHandler class implements three methods: receive, emit and copy. The receive method is called when a new frame is received from the client, and the emit method returns the next frame to send to the client. The copy method is called at the beginning of the stream to ensure each user has a unique stream handler.
  2. The emit method SHOULD NOT block. If a frame is not ready to be sent, the method should return None.

Async Stream Handlers

It is also possible to create asynchronous stream handlers. This is very convenient for accessing async APIs from major LLM developers, like Google and OpenAI. The main difference is that receive and emit are now defined with async def.

Here is a complete example of using AsyncStreamHandler for using the Google Gemini real time API:

AsyncStreamHandler
import asyncio
import base64
import logging
import os

import gradio as gr
import numpy as np
from google import genai
from gradio_webrtc import (
    AsyncStreamHandler,
    WebRTC,
    async_aggregate_bytes_to_16bit,
    get_twilio_turn_credentials,
)

class GeminiHandler(AsyncStreamHandler):
    def __init__(
        self, expected_layout="mono", output_sample_rate=24000, output_frame_size=480
    ) -> None:
        super().__init__(
            expected_layout,
            output_sample_rate,
            output_frame_size,
            input_sample_rate=16000,
        )
        self.client: genai.Client | None = None
        self.input_queue = asyncio.Queue()
        self.output_queue = asyncio.Queue()
        self.quit = asyncio.Event()

    def copy(self) -> "GeminiHandler":
        return GeminiHandler(
            expected_layout=self.expected_layout,
            output_sample_rate=self.output_sample_rate,
            output_frame_size=self.output_frame_size,
        )

    async def stream(self):
        while not self.quit.is_set():
            audio = await self.input_queue.get()
            yield audio

    async def connect(self, api_key: str):
        client = genai.Client(api_key=api_key, http_options={"api_version": "v1alpha"})
        config = {"response_modalities": ["AUDIO"]}
        async with client.aio.live.connect(
            model="gemini-2.0-flash-exp", config=config
        ) as session:
            async for audio in session.start_stream(
                stream=self.stream(), mime_type="audio/pcm"
            ):
                if audio.data:
                    yield audio.data

    async def receive(self, frame: tuple[int, np.ndarray]) -> None:
        _, array = frame
        array = array.squeeze()
        audio_message = base64.b64encode(array.tobytes()).decode("UTF-8")
        self.input_queue.put_nowait(audio_message)

    async def generator(self):
        async for audio_response in async_aggregate_bytes_to_16bit(
            self.connect(api_key=self.latest_args[1])
        ):
            self.output_queue.put_nowait(audio_response)

    async def emit(self):
        if not self.args_set.is_set():
            await self.wait_for_args()
        asyncio.create_task(self.generator())

        array = await self.output_queue.get()
        return (self.output_sample_rate, array)

    def shutdown(self) -> None:
        self.quit.set()

with gr.Blocks() as demo:
    gr.HTML(
        """
        <div style='text-align: center'>
            <h1>Gen AI SDK Voice Chat</h1>
            <p>Speak with Gemini using real-time audio streaming</p>
            <p>Get an API Key <a href="https://support.google.com/googleapi/answer/6158862?hl=en">here</a></p>
        </div>
    """
    )
    with gr.Row() as api_key_row:
        api_key = gr.Textbox(
            label="API Key",
            placeholder="Enter your API Key",
            value=os.getenv("GOOGLE_API_KEY", ""),
            type="password",
        )
    with gr.Row(visible=False) as row:
        webrtc = WebRTC(
            label="Audio",
            modality="audio",
            mode="send-receive",
            rtc_configuration=get_twilio_turn_credentials(),
            pulse_color="rgb(35, 157, 225)",
            icon_button_color="rgb(35, 157, 225)",
            icon="https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png",
        )

    webrtc.stream(
        GeminiHandler(),
        inputs=[webrtc, api_key],
        outputs=[webrtc],
        time_limit=90,
        concurrency_limit=2,
    )
    api_key.submit(
        lambda: (gr.update(visible=False), gr.update(visible=True)),
        None,
        [api_key_row, row],
    )

demo.launch()

Accessing Other Component Values from a StreamHandler

In the gemini demo above, you'll notice that we have the user input their google API key. This is stored in a gr.Textbox parameter. We can access the value of this component via the latest_args prop of the StreamHandler. The latest_args is a list storing the values of each component in the WebRTC stream event inputs parameter. The value of the WebRTC component is the 0th index and it's always the dummy string __webrtc_value__.

In order to fetch the latest value from the user however, we await self.wait_for_args(). In a synchronous StreamHandler, we would call self.wait_for_args_sync().

Server-To-Client Only

To stream only from the server to the client, implement a python generator and pass it to the component's stream event. The stream event must also specify a trigger corresponding to a UI interaction that starts the stream. In this case, it's a button click.

Server-To-CLient
import gradio as gr
from gradio_webrtc import WebRTC
from pydub import AudioSegment

def generation(num_steps):
    for _ in range(num_steps):
        segment = AudioSegment.from_file("audio_file.wav")
        array = np.array(segment.get_array_of_samples()).reshape(1, -1)
        yield (segment.frame_rate, array)

with gr.Blocks() as demo:
    audio = WebRTC(label="Stream", mode="receive",  # (1)
                   modality="audio")
    num_steps = gr.Slider(label="Number of Steps", minimum=1,
                          maximum=10, step=1, value=5)
    button = gr.Button("Generate")

    audio.stream(
        fn=generation, inputs=[num_steps], outputs=[audio],
        trigger=button.click # (2)
    )
  1. Set mode="receive" to only receive audio from the server.
  2. The stream event must take a trigger that corresponds to the gradio event that starts the stream. In this case, it's the button click.
  1. Set mode="receive" to only receive audio from the server.
  2. The stream event must take a trigger that corresponds to the gradio event that starts the stream. In this case, it's the button click.

Video Streaming

Input/Output Streaming

Set up a video Input/Output stream to continuosly receive webcam frames from the user and run an arbitrary python function to return a modified frame.

Input/Output Streaming
import gradio as gr
from gradio_webrtc import WebRTC


def detection(image, conf_threshold=0.3): # (1)
    ... your detection code here ...
    return modified_frame # (2)


with gr.Blocks() as demo:
    image = WebRTC(label="Stream", mode="send-receive", modality="video") # (3)
    conf_threshold = gr.Slider(
        label="Confidence Threshold",
        minimum=0.0,
        maximum=1.0,
        step=0.05,
        value=0.30,
    )
    image.stream(
        fn=detection,
        inputs=[image, conf_threshold], # (4)
        outputs=[image], time_limit=10
    )

if __name__ == "__main__":
    demo.launch()
  1. The webcam frame will be represented as a numpy array of shape (height, width, RGB).
  2. The function must return a numpy array. It can take arbitrary values from other components.
  3. Set the modality="video" and mode="send-receive"
  4. The inputs parameter should be a list where the first element is the WebRTC component. The only output allowed is the WebRTC component.
  1. The webcam frame will be represented as a numpy array of shape (height, width, RGB).
  2. The function must return a numpy array. It can take arbitrary values from other components.
  3. Set the modality="video" and mode="send-receive"
  4. The inputs parameter should be a list where the first element is the WebRTC component. The only output allowed is the WebRTC component.

Server-to-Client Only

Set up a server-to-client stream to stream video from an arbitrary user interaction.

Server-To-Client
    import gradio as gr
    from gradio_webrtc import WebRTC
    import cv2

    def generation():
        url = "https://download.tsi.telecom-paristech.fr/gpac/dataset/dash/uhd/mux_sources/hevcds_720p30_2M.mp4"
        cap = cv2.VideoCapture(url)
        iterating = True
        while iterating:
            iterating, frame = cap.read()
            yield frame # (1)

    with gr.Blocks() as demo:
        output_video = WebRTC(label="Video Stream", mode="receive", # (2)
                              modality="video")
        button = gr.Button("Start", variant="primary")
        output_video.stream(
            fn=generation, inputs=None, outputs=[output_video],
            trigger=button.click # (3)
        )
        demo.launch()
  1. The stream event's fn parameter is a generator function that yields the next frame from the video as a numpy array.
  2. Set mode="receive" to only receive audio from the server.
  3. The trigger parameter the gradio event that will trigger the stream. In this case, the button click event.
  1. The stream event's fn parameter is a generator function that yields the next frame from the video as a numpy array.
  2. Set mode="receive" to only receive audio from the server.
  3. The trigger parameter the gradio event that will trigger the stream. In this case, the button click event.

Audio-Video Streaming

You can simultaneously stream audio and video simultaneously to/from a server using AudioVideoStreamHandler or AsyncAudioVideoStreamHandler. They are identical to the audio StreamHandlers with the addition of video_receive and video_emit methods which take and return a numpy array, respectively.

Here is an example of the video handling functions for connecting with the Gemini multimodal API. In this case, we simply reflect the webcam feed back to the user but every second we'll send the latest webcam frame (and an additional image component) to the Gemini server.

Please see the "Gemini Audio Video Chat" example in the cookbook for the complete code.

Async Gemini Video Handling
async def video_receive(self, frame: np.ndarray):
    """Send video frames to the server"""
    if self.session:
        # send image every 1 second
        # otherwise we flood the API
        if time.time() - self.last_frame_time > 1:
            self.last_frame_time = time.time()
            await self.session.send(encode_image(frame))
            if self.latest_args[2] is not None:
                await self.session.send(encode_image(self.latest_args[2]))
    self.video_queue.put_nowait(frame)

async def video_emit(self) -> VideoEmitType:
    """Return video frames to the client"""
    return await self.video_queue.get()

Additional Outputs

In order to modify other components from within the WebRTC stream, you must yield an instance of AdditionalOutputs and add an on_additional_outputs event to the WebRTC component.

This is common for displaying a multimodal text/audio conversation in a Chatbot UI.

Additional Outputs
from gradio_webrtc import AdditionalOutputs, WebRTC

def transcribe(audio: tuple[int, np.ndarray],
               transformers_convo: list[dict],
               gradio_convo: list[dict]):
    response = model.generate(**inputs, max_length=256)
    transformers_convo.append({"role": "assistant", "content": response})
    gradio_convo.append({"role": "assistant", "content": response})
    yield AdditionalOutputs(transformers_convo, gradio_convo) # (1)


with gr.Blocks() as demo:
    gr.HTML(
    """
    <h1 style='text-align: center'>
    Talk to Qwen2Audio (Powered by WebRTC ⚡️)
    </h1>
    """
    )
    transformers_convo = gr.State(value=[])
    with gr.Row():
        with gr.Column():
            audio = WebRTC(
                label="Stream",
                mode="send", # (2)
                modality="audio",
            )
        with gr.Column():
            transcript = gr.Chatbot(label="transcript", type="messages")

    audio.stream(ReplyOnPause(transcribe),
                inputs=[audio, transformers_convo, transcript],
                outputs=[audio], time_limit=90)
    audio.on_additional_outputs(lambda s,a: (s,a), # (3)
                                outputs=[transformers_convo, transcript],
                                queue=False, show_progress="hidden")
    demo.launch()
  1. Pass your data to AdditionalOutputs and yield it.
  2. In this case, no audio is being returned, so we set mode="send". However, if we set mode="send-receive", we could also yield generated audio and AdditionalOutputs.
  3. The on_additional_outputs event does not take inputs. It's common practice to not run this event on the queue since it is just a quick UI update.
  1. Pass your data to AdditionalOutputs and yield it.
  2. In this case, no audio is being returned, so we set mode="send". However, if we set mode="send-receive", we could also yield generated audio and AdditionalOutputs.
  3. The on_additional_outputs event does not take inputs. It's common practice to not run this event on the queue since it is just a quick UI update.