Tired of “Hey Google” or “Alexa”?
Train your own French wake word (e.g., “Alexa”, “Jarvis”, “Lumière”) offline using OpenWakeWord + Piper TTS—directly in Google Colab!
No local setup needed (runs in the cloud) Supports French (with fr_FR-upmc-medium voice model) Export to .tflite for Home Assistant integration
Click “Show code” (under the first cell) to reveal the script.
Step 2: Replace the Code for French Support
Copy-paste this entire block to replace the default code: (The script downloads a French TTS model and generates samples with proper resampling.)
# @title { display-mode: "form" }
# @markdown # 1. Test Example Training Clip Generation
# @markdown Since openWakeWord models are trained on synthetic examples of your
# @markdown target wake word, it's a good idea to make sure that the examples
# @markdown sound correct. Type in your target wake word below, and run the
# @markdown cell to listen to it.
# @markdown
# @markdown Here are some tips that can help get the wake word to sound right:
# @markdown - If your wake word isn't being pronounced in the way
# @markdown you want, try spelling out the sounds phonetically with underscores
# @markdown separating each part.
# @markdown For example: "hey siri" --> "hey_seer_e".
# @markdown - Spell out numbers ("2" --> "two")
# @markdown - Avoid all punctuation except for "?" and "!", and remove unicode characters
import os
import sys
from IPython.display import Audio
if not os.path.exists("./piper-sample-generator"):
!git clone https://github.com/rhasspy/piper-sample-generator
!wget -O piper-sample-generator/models/fr_FR-upmc-medium.onnx 'https://huggingface.co/rhasspy/piper-voices/resolve/main/fr/fr_FR/upmc/medium/fr_FR-upmc-medium.onnx'
!wget -O piper-sample-generator/models/fr_FR-upmc-medium.onnx.json 'https://huggingface.co/rhasspy/piper-voices/resolve/main/fr/fr_FR/upmc/medium/fr_FR-upmc-medium.onnx.json'
!cd piper-sample-generator
# Install system dependencies
!pip install piper-tts piper-phonemize-cross
!pip install webrtcvad
!pip install 'torch<=2.5' torchvision torchaudio
if "piper-sample-generator/" not in sys.path:
sys.path.append("piper-sample-generator/")
import generate_samples
target_word = 'alexa' # @param {type:"string"}
def text_to_speech(text):
generate_samples.generate_samples_onnx(text = text,
max_samples=1,
length_scales=[1.1],
noise_scales=[0.7], noise_scale_ws = [0.7],
output_dir = './', batch_size=1, auto_reduce_batch_size=True,
file_names=["test_generation.wav"],
model="piper-sample-generator/models/fr_FR-upmc-medium.onnx"
)
text_to_speech(target_word)
Audio("test_generation.wav", autoplay=True)
What this does:
Downloads the French Piper TTS model (fr_FR-upmc-medium).
Generates a test audio clip of your wake word.
Plays the clip automatically for preview.
Step 3: Test Your Wake Word
Replace 'alexa' in target_word with your desired wake word (e.g., 'jarvis', 'lumière').
Run the cell ( button).
Listen to the generated clip. If it sounds wrong:
Try phonetic spelling (e.g., 'j_a_r_v_i_s' for “Jarvis”).
Adjust length_scales or noise_scales for clearer audio.
Step 4: Modify the Training Script
In Colab’s file explorer ( icon on the left), navigate to:
openwakeword/openwakeword/train.py
Double-click to open the file, then replace its entire content with this optimized script(see below for key improvements).
Hello,
Thank you for this guide in French. I tried to launch the wake word creation process, but it just downloads the .onnx file for my wake word, with an error:
**Traceback (most recent call last):**
** File "/content/openwakeword/openwakeword/train.py", line 920, in <module>**
** convert_onnx_to_tflite(os.path.join(config["output_dir"], config["model_name"] + ".onnx"),**
** File "/content/openwakeword/openwakeword/train.py", line 576, in convert_onnx_to_tflite**
** from onnx_tf.backend import prepare**
**ModuleNotFoundError: No module named 'onnx_tf'**
**WARNING: All log messages before absl::InitializeLog() is called are written to STDERR**
**E0000 00:00:1765544520.960328 16238 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered**
**E0000 00:00:1765544520.971439 16238 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered**
**W0000 00:00:1765544520.991113 16238 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.**
**W0000 00:00:1765544520.991157 16238 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.**
**W0000 00:00:1765544520.991163 16238 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.**
**W0000 00:00:1765544520.991169 16238 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.**
**Traceback (most recent call last):**
** File "/usr/local/bin/onnx2tf", line 5, in <module>**
** from onnx2tf import main**
** File "/usr/local/lib/python3.12/dist-packages/onnx2tf/__init__.py", line 1, in <module>**
** from onnx2tf.onnx2tf import convert, main**
** File "/usr/local/lib/python3.12/dist-packages/onnx2tf/onnx2tf.py", line 39, in <module>**
** import onnx_graphsurgeon as gs**
** File "/usr/local/lib/python3.12/dist-packages/onnx_graphsurgeon/__init__.py", line 1, in <module>**
** from onnx_graphsurgeon.exporters.onnx_exporter import export_onnx**
** File "/usr/local/lib/python3.12/dist-packages/onnx_graphsurgeon/exporters/onnx_exporter.py", line 134, in <module>**
** np.uint16, onnx.helper.float32_to_bfloat16**
** ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^**
**AttributeError: module 'onnx.helper' has no attribute 'float32_to_bfloat16'**
**mv: cannot stat 'my_custom_model/Jarviss_float32.tflite': No such file or directory**
**---------------------------------------------------------------------------**
**FileNotFoundError Traceback (most recent call last)**
**/tmp/ipython-input-2129544665.py in <cell line: 0>()**
** 106 **
** 107 files.download(f"my_custom_model/{config['model_name']}.onnx")**
**--> 108 files.download(f"my_custom_model/{config['model_name']}.tflite")**
**/usr/local/lib/python3.12/dist-packages/google/colab/files.py in download(filename)**
** 231 if not _os.path.exists(filename):**
** 232 msg = 'Cannot find file: {}'.format(filename)**
**--> 233 raise FileNotFoundError(msg) # pylint: disable=undefined-variable**
** 234 **
** 235 comm_manager = _IPython.get_ipython().kernel.comm_manager**
**FileNotFoundError: Cannot find file: my_custom_model/Jarviss.tflite**
A big thank you to @Atlantice for their work on the Colab Notebook, which helped me successfully train my own model.
@BadGones0l, I identified the problem: the original version is no longer compatible due to recent TensorFlow updates from Google. To convert an ONNX model to TFLite, make sure to use these exact library versions:
Building on @Atlantice’s contributions, I’ve written a step-by-step tutorial and updated the Colab notebook. You can access it here: French tutorial on my blog
Note: The article is in French, but the updated Colab notebook is linked inside for direct use.
Thanks for your work, I successfully created my custom Wake Word.
I just have a few optimizations to better detect the wake word (either in my Satellite Wyoming or in the model’s false_activation_penalty).
Hi and thank you for your work on this french tutorial for wakeword!
I tried your colab netbook but I get some errors and the process stops when I run step 2, perhaps it’s another version compatibility problem?
I’ll try again an see if I can debug it.
Tired of “Hey Google” or “Alexa”?
No local setup needed (runs in the cloud)
Prerequisites
Step 1: Open the Colab Notebook
Step 2: Replace the Code for French Support
Step 3: Test Your Wake Word
button).
Step 4: Modify the Training Script
icon on the left), navigate to:
Note: The script automatically resamples audio to 16kHz. If you skip this, training will fail!
Next Steps (To-Do List)
Planned