audio-recorder-streamlit

Audio record streamlit

This streamlit component allows to register an audio utterence from a user.

Installation

pip install audio-recorder-streamlit

Usage

import streamlit as st
from audio_recorder_streamlit import audio_recorder

audio_bytes = audio_recorder()
if audio_bytes:
    st.audio(audio_bytes, format="audio/wav")

Recording parameters

You can adjust the recording parameters energy_threshold and pause_threshold:

• energy_threshold: The energy recording sensibility above which we consider that the user is speaking. If it is a float, then this is the energy threshold used to automatically detect recording start and recording end. You can provide a tuple for specifying different threshold for recording start detection and recording end detection.
• pause_threshold: The number of seconds to spend below energy_level to automatically stop the recording.
• sample_rate: Sample rate of the recorded audio. If not provided, this will use the default sample rate (https://developer.mozilla.org/en-US/docs/Web/API/AudioContext/AudioContext).
• auto_start: If True starts the recording as soon as the component has loaded.

# The recording will stop automatically
# 2 sec after the utterance end
audio_bytes = audio_recorder(pause_threshold=2.0, sample_rate=41_000)

Styling parameters

You can adjust the button style parameters:

• Text: specify the text displayed next to the icon
• Icon: specify the icon among the font awesome solid icons (https://fontawesome.com/search?o=r&s=solid)
• Color: specify the neutral color and recording color.
• Size: specify the icon size using font awesome sizing (https://fontawesome.com/docs/web/style/size).

audio_bytes = audio_recorder(
    text="",
    recording_color="#e8b62c",
    neutral_color="#6aa36f",
    icon_name="user",
    icon_size="6x",
)

Frequently Asked Question

How can I record for a fixed duration ?

You can record for a fixed duration by setting the energy_threshold=(-1.0, 1.0) so that the recorder considers that you are speaking at the beginning and then you are never speaking from this point on.

Then simply set pause_threshold to your desired recording length.

# Records 3 seconds in any case
audio_bytes = audio_recorder(
  energy_threshold=(-1.0, 1.0),
  pause_threshold=3.0,
)