Easy-to-use, low-latency speech-to-text library for realtime applications
Custom wake words with OpenWakeWord. Thanks to the developers of this!
RealtimeSTT listens to the microphone and transcribes voice into text.
Hint: Check out Linguflex, the original project from which RealtimeSTT is spun off. It lets you control your environment by speaking and is one of the most capable and sophisticated open-source assistants currently available.
It's ideal for:
https://github.com/KoljaB/RealtimeSTT/assets/7604638/207cb9a2-4482-48e7-9d2b-0722c3ee6d14
Latest Version: v0.2.1
See release history.
Hint: Since we use the
multiprocessingmodule now, ensure to include theif __name__ == '__main__':protection in your code to prevent unexpected behavior, especially on platforms like Windows. For a detailed explanation on why this is important, visit the official Python documentation onmultiprocessing.
Hint: Check out RealtimeTTS, the output counterpart of this library, for text-to-voice capabilities. Together, they form a powerful realtime audio wrapper around large language models.
This library uses:
These components represent the "industry standard" for cutting-edge applications, providing the most modern and effective foundation for building high-end solutions.
pip install RealtimeSTTThis will install all the necessary dependencies, including a CPU support only version of PyTorch.
Although it is possible to run RealtimeSTT with a CPU installation only (use a small model like "tiny" or "base" in this case) you will get way better experience using:
To upgrade your PyTorch installation to enable GPU support with CUDA, follow these instructions based on your specific CUDA version. This is useful if you wish to enhance the performance of RealtimeSTT with CUDA capabilities.
To update PyTorch and Torchaudio to support CUDA 11.8, use the following commands:
pip install torch==2.3.1+cu118 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118To update PyTorch and Torchaudio to support CUDA 12.X, execute the following:
pip install torch==2.3.1+cu121 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu121Replace 2.3.1 with the version of PyTorch that matches your system and requirements.
Note: To check if your NVIDIA GPU supports CUDA, visit the official CUDA GPUs list.
If you didn't use CUDA models before, some additional steps might be needed one time before installation. These steps prepare the system for CUDA support and installation of the GPU-optimized installation. This is recommended for those who require better performance and have a compatible NVIDIA GPU. To use RealtimeSTT with GPU support via CUDA please also follow these steps:
Install NVIDIA CUDA Toolkit:
Install NVIDIA cuDNN:
Install ffmpeg:
Note: Installation of ffmpeg might not actually be needed to operate RealtimeSTT *thanks to jgilbert2017 for pointing this out
You can download an installer for your OS from the ffmpeg Website.
Or use a package manager:
On Ubuntu or Debian:
sudo apt update && sudo apt install ffmpegOn Arch Linux:
sudo pacman -S ffmpegOn MacOS using Homebrew (https://brew.sh/):
brew install ffmpegOn Windows using Winget official documentation :
winget install Gyan.FFmpegOn Windows using Chocolatey (https://chocolatey.org/):
choco install ffmpegOn Windows using Scoop (https://scoop.sh/):
scoop install ffmpegBasic usage:
Start and stop of recording are manually triggered.
recorder.start()
recorder.stop()
print(recorder.text())Recording based on voice activity detection.
with AudioToTextRecorder() as recorder:
print(recorder.text())When running recorder.text in a loop it is recommended to use a callback, allowing the transcription to be run asynchronously:
def process_text(text):
print (text)
while True:
recorder.text(process_text)Keyword activation before detecting voice. Write the comma-separated list of your desired activation keywords into the wake_words parameter. You can choose wake words from these list: alexa, americano, blueberry, bumblebee, computer, grapefruits, grasshopper, hey google, hey siri, jarvis, ok google, picovoice, porcupine, terminator.
recorder = AudioToTextRecorder(wake_words="jarvis")
print('Say "Jarvis" then speak.')
print(recorder.text())You can set callback functions to be executed on different events (see Configuration) :
def my_start_callback():
print("Recording started!")
def my_stop_callback():
print("Recording stopped!")
recorder = AudioToTextRecorder(on_recording_start=my_start_callback,
on_recording_stop=my_stop_callback)If you don't want to use the local microphone set use_microphone parameter to false and provide raw PCM audiochunks in 16-bit mono (samplerate 16000) with this method:
recorder.feed_audio(audio_chunk)You can shutdown the recorder safely by using the context manager protocol:
with AudioToTextRecorder() as recorder:
[...]Or you can call the shutdown method manually (if using "with" is not feasible):
recorder.shutdown()The test subdirectory contains a set of scripts to help you evaluate and understand the capabilities of the RealtimeTTS library.
Test scripts depending on RealtimeTTS library may require you to enter your azure service region within the script. When using OpenAI-, Azure- or Elevenlabs-related demo scripts the API Keys should be provided in the environment variables OPENAI_API_KEY, AZURE_SPEECH_KEY and ELEVENLABS_API_KEY (see RealtimeTTS)
simple_test.py
realtimestt_test.py
wakeword_test.py
translator.py
pip install openai realtimetts.openai_voice_interface.py
pip install openai realtimetts.advanced_talk.py
pip install openai keyboard realtimetts.minimalistic_talkbot.py
pip install openai realtimetts.The example_app subdirectory contains a polished user interface application for the OpenAI API based on PyQt5.
AudioToTextRecorderWhen you initialize the AudioToTextRecorder class, you have various options to customize its behavior.
model (str, default="tiny"): Model size or path for transcription.
language (str, default=""): Language code for transcription. If left empty, the model will try to auto-detect the language. Supported language codes are listed in Whisper Tokenizer library.
compute_type (str, default="default"): Specifies the type of computation to be used for transcription. See Whisper Quantization
input_device_index (int, default=0): Audio Input Device Index to use.
gpu_device_index (int, default=0): GPU Device Index to use. The model can also be loaded on multiple GPUs by passing a list of IDs (e.g. [0, 1, 2, 3]).
device (str, default="cuda"): Device for model to use. Can either be "cuda" or "cpu".
on_recording_start: A callable function triggered when recording starts.
on_recording_stop: A callable function triggered when recording ends.
on_transcription_start: A callable function triggered when transcription starts.
ensure_sentence_starting_uppercase (bool, default=True): Ensures that every sentence detected by the algorithm starts with an uppercase letter.
ensure_sentence_ends_with_period (bool, default=True): Ensures that every sentence that doesn't end with punctuation such as "?", "!" ends with a period
use_microphone (bool, default=True): Usage of local microphone for transcription. Set to False if you want to provide chunks with feed_audio method.
spinner (bool, default=True): Provides a spinner animation text with information about the current recorder state.
level (int, default=logging.WARNING): Logging level.
handle_buffer_overflow (bool, default=True): If set, the system will log a warning when an input overflow occurs during recording and remove the data from the buffer.
beam_size (int, default=5): The beam size to use for beam search decoding.
initial_prompt (str or iterable of int, default=None): Initial prompt to be fed to the transcription models.
suppress_tokens (list of int, default=[-1]): Tokens to be suppressed from the transcription output.
on_recorded_chunk: A callback function that is triggered when a chunk of audio is recorded. Submits the chunk data as parameter.
debug_mode (bool, default=False): If set, the system prints additional debug information to the console.
Note: When enabling realtime description a GPU installation is strongly advised. Using realtime transcription may create high GPU loads.
enable_realtime_transcription (bool, default=False): Enables or disables real-time transcription of audio. When set to True, the audio will be transcribed continuously as it is being recorded.
realtime_model_type (str, default="tiny"): Specifies the size or path of the machine learning model to be used for real-time transcription.
realtime_processing_pause (float, default=0.2): Specifies the time interval in seconds after a chunk of audio gets transcribed. Lower values will result in more "real-time" (frequent) transcription updates but may increase computational load.
on_realtime_transcription_update: A callback function that is triggered whenever there's an update in the real-time transcription. The function is called with the newly transcribed text as its argument.
on_realtime_transcription_stabilized: A callback function that is triggered whenever there's an update in the real-time transcription and returns a higher quality, stabilized text as its argument.
beam_size_realtime (int, default=3): The beam size to use for real-time transcription beam search decoding.
silero_sensitivity (float, default=0.6): Sensitivity for Silero's voice activity detection ranging from 0 (least sensitive) to 1 (most sensitive). Default is 0.6.
silero_use_onnx (bool, default=False): Enables usage of the pre-trained model from Silero in the ONNX (Open Neural Network Exchange) format instead of the PyTorch format. Default is False. Recommended for faster performance.
webrtc_sensitivity (int, default=3): Sensitivity for the WebRTC Voice Activity Detection engine ranging from 0 (least aggressive / most sensitive) to 3 (most aggressive, least sensitive). Default is 3.
post_speech_silence_duration (float, default=0.2): Duration in seconds of silence that must follow speech before the recording is considered to be completed. This ensures that any brief pauses during speech don't prematurely end the recording.
min_gap_between_recordings (float, default=1.0): Specifies the minimum time interval in seconds that should exist between the end of one recording session and the beginning of another to prevent rapid consecutive recordings.
min_length_of_recording (float, default=1.0): Specifies the minimum duration in seconds that a recording session should last to ensure meaningful audio capture, preventing excessively short or fragmented recordings.
pre_recording_buffer_duration (float, default=0.2): The time span, in seconds, during which audio is buffered prior to formal recording. This helps counterbalancing the latency inherent in speech activity detection, ensuring no initial audio is missed.
on_vad_detect_start: A callable function triggered when the system starts to listen for voice activity.
on_vad_detect_stop: A callable function triggered when the system stops to listen for voice activity.
wakeword_backend (str, default="pvporcupine"): Specifies the backend library to use for wake word detection. Supported options include 'pvporcupine' for using the Porcupine wake word engine or 'oww' for using the OpenWakeWord engine.
openwakeword_model_paths (str, default=None): Comma-separated paths to model files for the openwakeword library. These paths point to custom models that can be used for wake word detection when the openwakeword library is selected as the wakeword_backend.
openwakeword_inference_framework (str, default="onnx"): Specifies the inference framework to use with the openwakeword library. Can be either 'onnx' for Open Neural Network Exchange format or 'tflite' for TensorFlow Lite.
wake_words (str, default=""): Initiate recording when using the 'pvporcupine' wakeword backend. Multiple wake words can be provided as a comma-separated string. Supported wake words are: alexa, americano, blueberry, bumblebee, computer, grapefruits, grasshopper, hey google, hey siri, jarvis, ok google, picovoice, porcupine, terminator. For the 'openwakeword' backend, wake words are automatically extracted from the provided model files, so specifying them here is not necessary.
wake_words_sensitivity (float, default=0.6): Sensitivity level for wake word detection (0 for least sensitive, 1 for most sensitive).
wake_word_activation_delay (float, default=0): Duration in seconds after the start of monitoring before the system switches to wake word activation if no voice is initially detected. If set to zero, the system uses wake word activation immediately.
wake_word_timeout (float, default=5): Duration in seconds after a wake word is recognized. If no subsequent voice activity is detected within this window, the system transitions back to an inactive state, awaiting the next wake word or voice activation.
wake_word_buffer_duration (float, default=0.1): Duration in seconds to buffer audio data during wake word detection. This helps in cutting out the wake word from the recording buffer so it does not falsely get detected along with the following spoken text, ensuring cleaner and more accurate transcription start triggers. Increase this if parts of the wake word get detected as text.
on_wakeword_detected: A callable function triggered when a wake word is detected.
on_wakeword_timeout: A callable function triggered when the system goes back to an inactive state after when no speech was detected after wake word activation.
on_wakeword_detection_start: A callable function triggered when the system starts to listen for wake words
on_wakeword_detection_end: A callable function triggered when stopping to listen for wake words (e.g. because of timeout or wake word detected)
Look here for information about how to train your own OpenWakeWord models. You can use a simple Google Colab notebook for a start or use a more detailed notebook that enables more customization (can produce high quality models, but requires more development experience).
You might need to use tf2onnx to convert tensorflow tflite models to onnx format:
pip install -U tf2onnx
python -m tf2onnx.convert --tflite my_model_filename.tflite --output my_model_filename.onnxSuggested starting parameters for OpenWakeWord usage:
with AudioToTextRecorder(
wakeword_backend="oww",
wake_words_sensitivity=0.35,
openwakeword_model_paths="word1.onnx,word2.onnx",
wake_word_buffer_duration=1,
) as recorder:Contributions are always welcome!
Shoutout to Steven Linn for providing docker support.
MIT
Kolja Beigel
Email: kolja.beigel@web.de
GitHub