blog.matterxiaomi.com

Matter Bridge in Home Assistant Part2 - Install MatterBridge Connect to Home assistant

blog.matterxiaomi.com — Sun, 15 Feb 2026 00:02:19 GMT

Matter Bridge in Home Assistant Part2 - Install MatterBridge Connect to Home assistant

Quick start
Install and configure
How to Use

Quick start

I set up the matterbridge as follows:

Install the Matterbridge docker

Create long-lived access tokens to allow home-assistant-matter-hub docker to interact with your Home Assistant instance.

Communication configure between Matterhub and Home Assistant,Matterbridge to connect to home assistant with url and token

expose homeassistant device as a matter bridge

open http://192.168.2.125:8482/ via chrome browser

Create a new bridge,

Add device "pattern: switch.air_con" in new bridge

start it to generate a pairing QR code

Connect accessory to Apple Home

Install and configure

docker-compose.yml

You need to create an access token in home assistant instance and export it like this:

services:
  matter-hub:
    image: ghcr.io/t0bst4r/home-assistant-matter-hub:3.0.1
    restart: unless-stopped
    network_mode: host
    environment: # more options can be found in the configuration section
      - HAMH_HOME_ASSISTANT_URL=http://192.168.2.125:8123/
      - HAMH_HOME_ASSISTANT_ACCESS_TOKEN=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJhMzcwZDExYjM4MjE0YzFmYThmZTk3NDZjMDQyODU2NSIsImlhdCI6MTc3MTA4MTk1NSwiZXhwIjoyMDg2NDQxOTU1fQ.vhZD-KhJe4XIXd6_XvBE92y4T5W1aICSfBCbTTCvFL4
      - HAMH_LOG_LEVEL=info
      - HAMH_HTTP_PORT=8482
    volumes:
      - /datadocker/home-assistant-matter-hub:/data

Now you can visit it via web ui

http://192.168.2.125:8482/

How to Use

expose homeassistant device as a matter bridge

open http://192.168.2.125:8482/

Create a new bridge for device,home-assistant-matter-hub docker get it from home assistant via api.

type:pattern

value:light.yeelink_cn_ceiling21_s_2_light

{
  "name": "matterbridgeceilling21v2",
  "port": 5543,
  "filter": {
    "include": [
      {
        "type": "pattern",
        "value": "light.yeelink_cn_476690814_ceiling21_s_2_light"
      }
    ],
    "exclude": []
  },
  "featureFlags": {
    "coverDoNotInvertPercentage": false,
    "includeHiddenEntities": false
  }
}

Ecovacs in Home Assistant Part6 - Robot Vacuum Control MCP Server

blog.matterxiaomi.com — Fri, 13 Feb 2026 13:17:29 GMT

MCP protocol

https://github.com/ecovacs-ai/ecovacs-mcp/blob/main/ecovacs_mcp/robot_mcp_stdio.py

custom integration

https://github.com/hoangminh1109/ecovacs_cn

https://open.ecovacs.com/#/serviceOverview

ModelScope vs Hugging Face vs k2-fsa.github.io vs Kaldi vs Sherpa

blog.matterxiaomi.com — Wed, 11 Feb 2026 19:46:43 GMT

Hugging Face, ModelScope, and k2-fsa.github.io (specifically the k2-fsa/sherpa-onnx project) represent different approaches to the machine learning ecosystem。

Hugging Face and ModelScope host everything. k2-fsa and Sherpa only do Speech.

k2-fsa and Sherpa are highly specialized tools focused on speech recognition (ASR) and synthesis (TTS)

Sherpa (often referred to as sherpa-onnx or sherpa-ncnn) is a lightweight speech-to-text (ASR) and text-to-speech (TTS) engine. Best for: Deploying speech models on edge devices (Android, iOS, WebAssembly, ARM boards) or high-performance servers, prioritizing low latency and CPU efficiency.

Hugging Face（Global AI model platform）
ModelScope(The Alibaba/Chinese AI Industrial model platform)
specialized tools focused on speech recognition (ASR) and synthesis (TTS)
- k2-fsa (Next-Gen Kaldi)
- Sherpa(The Real-Time Speech Deployment Tool)

These four entities represent two different categories: Model Ecosystems (Hugging Face & ModelScope) and Speech Recognition Frameworks (Kaldi & k2-fsa).

Hugging Face（Global AI model platform）

The Global Industry Standard。It supports NLP, computer vision, audio, and multimodal models via transformers and diffusers libraries.

download models

https://huggingface.co/models

https://huggingface.co/FunAudioLLM/SenseVoiceSmall

https://huggingface.co/funasr/paraformer-zh/blame/7904416f6cb6290ee7dc0b2ddb2993a9fe4f421a/README.md

ModelScope(The Alibaba/Chinese AI Industrial model platform)

ModelScope is an AI model hub led by Alibaba DAMO Academy. It provides pre-trained models, pipelines, and deployment tools, especially strong in Chinese language and speech technologies.ModelScope is often described as the "Chinese Hugging Face."

https://modelscope.cn/search?search=sherpa%20onnx

Inference Framework

1.funasr

2.funasr-onnx

see:https://github.com/modelscope/FunASR?tab=readme-ov-file#sensevoice

specialized tools focused on speech recognition (ASR) and synthesis (TTS)

Kaldi: Speech Toolkit

The "grandfather" of modern speech recognition. It is a C++ based toolkit developed primarily by Dan Povey.

demo

KaldiRecognizer

https://github.com/rhasspy/wyoming-faster-whisper/blob/main/wyoming_faster_whisper/__main__.py

k2-fsa (Next-Gen Kaldi)

Speech Toolkit.The Modern Successor

What it is: Often called "Next-gen Kaldi." It is a complete rewrite of Kaldi’s core concepts to make them natively compatible with PyTorch.

Key Repositories:

Icefall: Where the actual training recipes for speech models (like Zipformer) live.

k2: The core library for differentiable FSTs.the old version of the tool you used before k2-fsa existed.

download sherpa-onnx

https://github.com/k2-fsa/sherpa-onnx/releases

download sherpa-onnx asr models

https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html

https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models

download Silero VAD ONNX model

# https://k2-fsa.github.io/sherpa/onnx/sense-voice/pretrained.html#sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx

download url:https://k2-fsa.github.io/sherpa/onnx/vad/silero-vad.html#download-models-files

Sherpa(The Real-Time Speech Deployment Tool)

The deployment engine (CPU/GPU, Android, iOS, WebAssembly).It uses models trained in the k2-fsa ecosystem.

How they all fit together

1.k2-fsa is the tool you use to build a high-performance speech model.

silero-vad model download

https://k2-fsa.github.io/sherpa/onnx/vad/silero-vad.html#download-models-files

Once you've trained that model using k2-fsa, you might upload it to Hugging Face or ModelScope so others can download it easily.

Hugging Face hosts models from both ModelScope and k2-fsa/Sherpa, serving as a distribution point for them.

Layer Relationship

Model Hosting & Distribution
   ├── ModelScope
   └── Hugging Face

Inference / Runtime Framework
   └── k2-fsa / sherpa

Create Wyoming server for Home assistant Part2 - stt - wyoming-funasr arm64

blog.matterxiaomi.com — Thu, 05 Feb 2026 17:55:26 GMT

Wyoming protocol server for the funasr speech to text system.stt - wyoming-funasr arm64

FunASR: A Fundamental End-to-End Speech Recognition Toolkit.

To make an STT server work with Home Assistant, the industry standard is using the Wyoming Protocol.

Step 1.Development Environment Setup
- Create Python virtual environment
Step 2. Install
Step3.Download and test a model (example: paraformer-zh)
- SenseVoice - Speech Recognition (Non-streaming)
Step 4.FunASR + Wyoming STT full server
- Install wyoming
Strategies to reduce latency

Step 1.Development Environment Setup

Create Python virtual environment

mkdir -p /funasr-wyoming

cd /funasr-wyoming

python3 -m venv venv

source venv/bin/activate

python --version

Python 3.11.2

pip3 install wyoming==1.8.0

pip3 install funasr==1.3.0

pip3 install torch

pip3 install torchaudio

apt list --installed

(venv) root@raspberrypi:/funasr-wyoming# pip3 show funasr
Name: funasr
Version: 1.3.0
Summary: FunASR: A Fundamental End-to-End Speech Recognition Toolkit
Home-page: https://github.com/alibaba-damo-academy/FunASR.git
Author: Speech Lab of Alibaba Group
Author-email: funasr@list.alibaba-inc.com
License: The MIT License
Location: /funasr-wyoming/venv/lib/python3.11/site-packages
Requires: editdistance, hydra-core, jaconv, jamo, jieba, kaldiio, librosa, modelscope, oss2, pytorch_wpe, PyYAML, requests, scipy, sentencepiece, soundfile, tensorboardX, torch_complex, tqdm, umap_learn

Requirements

python>=3.8
torch>=1.13
torchaudio

Step 2. Install

(venv) root@raspberrypi:/funasr-wyoming# pip3 --version

pip 23.0.1 from /funasr-wyoming/venv/lib/python3.11/site-packages/pip (python 3.11)

Install torch via PyPI

pip3 install torch==2.1.0 (CPU-only)

output

Installing collected packages: mpmath, sympy, networkx, MarkupSafe, fsspec, jinja2, torch
Successfully installed MarkupSafe-3.0.3 fsspec-2026.1.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 sympy-1.14.0 torch-2.1.0

if ffmpeg is not installed. torchaudio is used to load audio

pip3 install torchaudio==2.1.0 (CPU-only)

output

Successfully installed torchaudio-2.1.0

You will need the wyoming and funasr libraries.

Install FunASR 1.3.0 via PyPI

pip3 install -U funasr==1.3.0

This will pull:

Downloading https://www.piwheels.org/simple/threadpoolctl/threadpoolctl-3.6.0-py3-none-any.whl (18 kB)

Installing collected packages: jieba, jamo, jaconv, crcmod, antlr4-python3-runtime, urllib3, typing_extensions, tqdm, threadpoolctl, six, sentencepiece, PyYAML, pycryptodome, pycparser, protobuf, platformdirs, packaging, numpy, msgpack, llvmlite, joblib, jmespath, idna, filelock, editdistance, decorator, charset_normalizer, certifi, audioread, torch_complex, tensorboardX, soxr, scipy, requests, pytorch_wpe, omegaconf, numba, lazy_loader, kaldiio, cffi, soundfile, scikit-learn, pooch, modelscope, hydra-core, cryptography, pynndescent, librosa, aliyun-python-sdk-core, umap_learn, aliyun-python-sdk-kms, oss2, funasr

output

Successfully installed PyYAML-6.0.3 aliyun-python-sdk-core-2.16.0 aliyun-python-sdk-kms-2.16.5 antlr4-python3-runtime-4.9.3 audioread-3.1.0 certifi-2026.1.4 cffi-2.0.0 charset_normalizer-3.4.4 crcmod-1.7 cryptography-46.0.3 decorator-5.2.1 editdistance-0.8.1 filelock-3.20.3 funasr-1.3.0 hydra-core-1.3.2 idna-3.11 jaconv-0.4.1 jamo-0.4.1 jieba-0.42.1 jmespath-0.10.0 joblib-1.5.3 kaldiio-2.18.1 lazy_loader-0.4 librosa-0.11.0 llvmlite-0.46.0 modelscope-1.34.0 msgpack-1.1.2 numba-0.63.1 numpy-2.3.5 omegaconf-2.3.0 oss2-2.19.1 packaging-26.0 platformdirs-4.5.1 pooch-1.8.2 protobuf-6.33.4 pycparser-3.0 pycryptodome-3.23.0 pynndescent-0.6.0 pytorch_wpe-0.0.1 requests-2.32.5 scikit-learn-1.8.0 scipy-1.17.0 sentencepiece-0.2.1 six-1.17.0 soundfile-0.13.1 soxr-1.0.0 tensorboardX-2.6.4 threadpoolctl-3.6.0 torch_complex-0.4.4 tqdm-4.67.1 typing_extensions-4.15.0 umap_learn-0.5.11 urllib3-2.6.3

detail:https://pypi.org/project/funasr

sudo apt install ffmpeg

output

ffmpeg is already the newest version (8:5.1.8-0+deb12u1+rpt1).

if ffmpeg is not installed. torchaudio is used to load audio

Verify installation

python - << 'EOF'
from funasr import AutoModel
print("FunASR imported OK")
EOF

output

FunASR imported OK

Step3.Download and test a model (example: paraformer-zh)

test.py

from funasr import AutoModel

model = AutoModel(
    model="paraformer-zh",
    model_revision="v2.0.4",
    vad_model="fsmn-vad",
    vad_model_revision="v2.0.4",
    punc_model="ct-punc",
    punc_model_revision="v2.0.4",
)

res = model.generate(input="test.wav")
print(res)

res = model.generate(input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav")
print(res)

SenseVoice - Speech Recognition (Non-streaming)

from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess

model_dir = "iic/SenseVoiceSmall"

model = AutoModel(
    model=model_dir,
    vad_model="fsmn-vad",
    vad_kwargs={"max_single_segment_time": 30000},
    device="cuda:0",
)

# en
res = model.generate(
    input=f"{model.model_path}/example/en.mp3",
    cache={},
    language="auto",  # "zn", "en", "yue", "ja", "ko", "nospeech"
    use_itn=True,
    batch_size_s=60,
    merge_vad=True,  #
    merge_length_s=15,
)
text = rich_transcription_postprocess(res[0]["text"])
print(text)

see:https://github.com/modelscope/FunASR?tab=readme-ov-file#sensevoice

python3 test.py

Models are cached in:

/root/.cache/modelscope/hub/models/iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch

output

Downloading Model from https://www.modelscope.cn to directory: /root/.cache/modelscope/hub/models/iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch


2026-01-25 08:58:28,492 - modelscope - INFO - Use user-specified model revision: v2.0.4
2026-01-25 08:58:28,595 - modelscope - INFO - Got 11 files, start to download ...
Downloading [fig/res.png]: 100%|███████████████████████████████████████████████████| 192k/192k [00:00<00:00, 386kB/s]
Downloading [am.mvn]: 100%|█████████████████████████████████████████████████████| 10.9k/10.9k [00:00<00:00, 21.7kB/s]
Downloading [example/hotword.txt]: 100%|███████████████████████████████████████████| 7.00/7.00 [00:00<00:00, 11.9B/s]
Downloading [config.yaml]: 100%|████████████████████████████████████████████████| 3.34k/3.34k [00:00<00:00, 5.66kB/s]
Downloading [configuration.json]: 100%|███████████████████████████████████████████████| 478/478 [00:00<00:00, 766B/s]
Downloading [README.md]: 100%|██████████████████████████████████████████████████| 11.3k/11.3k [00:00<00:00, 18.2kB/s]
Downloading [example/asr_example.wav]: 100%|███████████████████████████████████████| 141k/141k [00:00<00:00, 208kB/s]
Downloading [fig/seaco.png]: 100%|█████████████████████████████████████████████████| 167k/167k [00:00<00:00, 296kB/s]
Downloading [tokens.json]: 100%|█████████████████████████████████████████████████| 91.5k/91.5k [00:00<00:00, 165kB/s]
Downloading [seg_dict]: 100%|███████████████████████████████████████████████████| 7.90M/7.90M [00:03<00:00, 2.76MB/s]
Downloading [model.pt]: 100%|█████████████████████████████████████████████████████| 944M/944M [01:31<00:00, 10.8MB/s]
Processing 11 items: 100%|████████████████████████████████████████████████████████| 11.0/11.0 [01:31<00:00, 8.34s/it]
2026-01-25 09:00:00,347 - modelscope - INFO - Download model 'iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch' successfully.█████████████████████████████████████████████| 91.5k/91.5k [00:00<00:00, 165kB/s]
WARNING:root:trust_remote_code: False                                            | 5.00M/944M [00:01<02:50, 5.77MB/s]
Downloading [model.pt]:   2%|█▎                                                  | 23.0M/944M [00:03<02:15, 7.12MB/s]
Downloading [model.pt]: 100%|████████████████████████████████████████████████████▉| 942M/944M [01:31<00:00, 6.28MB/s]
Downloading [seg_dict]: 100%|███████████████████████████████████████████████████| 7.90M/7.90M [00:03<00:00, 4.16MB/s

python3 -c "from funasr import AutoModel; AutoModel(model='paraformer-zh', device='cpu')"

Step 4.FunASR + Wyoming STT full server

Install wyoming

pip3 install wyoming==1.8.0

output

Looking in indexes: https://pypi.org/simple, https://www.piwheels.org/simple
Collecting wyoming==1.5.0
  Downloading wyoming-1.5.0-py3-none-any.whl (23 kB)
Installing collected packages: wyoming
Successfully installed wyoming-1.5.0

A Wyoming server consists of an AsyncServer and an AsyncEventHandler. The handler processes events like Describe

python3 server.py

You will need the wyoming and funasr libraries.

Describe event

Listen for a Describe event (to tell HA it's an STT service)

AudioStart event

Ha client->

{
  "type": "audio.start",
  "rate": 16000,
  "width": 2,
  "channels": 1
}

AudioStart.is_type(event.type)

AudioChunk event

The AudioChunk event is where you collect the raw PCM data.Receive AudioChunk events and buffer them.

AudioStop event

Trigger the SenseVoice model when AudioStop event is received.AudioStop event is where you trigger the inference.

funasr-wyoming# python3 server.py
funasr version: 1.3.0.
Check update of funasr, and it would cost few times. You may disable it by set `disable_update=True` in AutoModel
New version is available: 1.3.1.
Please use the command "pip install -U funasr" to upgrade.
2026-02-19 14:15:51.321 - INFO - root - download models from model hub: ms
Downloading Model from https://www.modelscope.cn to directory: /root/.cache/modelscope/hub/models/iic/SenseVoiceSmall
2026-02-19 14:15:52.591 - WARNING - root - trust_remote_code: False
2026-02-19 14:15:54.679 - INFO - root - Loading pretrained params from /root/.cache/modelscope/hub/models/iic/SenseVoiceSmall/model.pt
2026-02-19 14:15:54.688 - INFO - root - ckpt: /root/.cache/modelscope/hub/models/iic/SenseVoiceSmall/model.pt
2026-02-19 14:16:10.752 - INFO - root - scope_map: ['module.', 'None']
2026-02-19 14:16:10.871 - INFO - root - excludes: None
2026-02-19 14:16:11.383 - INFO - root - Loading ckpt: /root/.cache/modelscope/hub/models/iic/SenseVoiceSmall/model.pt, status: <All keys matched successfully>
2026-02-19 14:16:11.461 - INFO - root - Building VAD model.
2026-02-19 14:16:11.461 - INFO - root - download models from model hub: ms
Downloading Model from https://www.modelscope.cn to directory: /root/.cache/modelscope/hub/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch
2026-02-19 14:16:13.308 - WARNING - root - trust_remote_code: False
2026-02-19 14:16:13.652 - INFO - root - Loading pretrained params from /root/.cache/modelscope/hub/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch/model.pt
2026-02-19 14:16:13.653 - INFO - root - ckpt: /root/.cache/modelscope/hub/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch/model.pt
2026-02-19 14:16:13.959 - INFO - root - scope_map: ['module.', 'None']
2026-02-19 14:16:13.959 - INFO - root - excludes: None
2026-02-19 14:16:13.962 - INFO - root - Loading ckpt: /root/.cache/modelscope/hub/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch/model.pt, status: <All keys matched successfully>
2026-02-19 14:16:14.020 - INFO - wyoming-funasr-stt-server - Wyoming STT Server started on port 10850
2026-02-19 14:16:19.596 - INFO - wyoming-funasr-stt-server - Transcribe received
2026-02-19 14:16:19.597 - INFO - wyoming-funasr-stt-server - AudioStart received
2026-02-19 14:16:22.530 - INFO - wyoming-funasr-stt-server - AudioStop received. Processing...  95040 bytes of audio...
2026-02-19 14:16:22.557 - 音频长度: 47520 samples, 2.97 秒
2026-02-19 14:16:22.559 - INFO - wyoming-funasr-stt-server - Audio length: 47520 samples, 2.97 s
rtf_avg: 0.578: 100%|█████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00,  1.71s/it]
rtf_avg: 1.793: 100%|█████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00,  1.72s/it]
rtf_avg: 0.581, time_speech:  2.970, time_escape: 1.727: 100%|████████████████████████████████████████████| 1/1 [00:01<00:00,  1.73s/it]
2026-02-19 14:16:26.056 - 过滤掉 SenseVoice 可能输出的情感/事件标签
2026-02-19 14:16:26.056 - INFO - wyoming-funasr-stt-server - 识别结果原文：可能输出的情感/事件标签- Result: <|zh|><|NEUTRAL|><|Speech|><|withitn|>换货。
2026-02-19 14:16:26.057 - INFO - wyoming-funasr-stt-server - Result: 换货。
2026-02-19 14:16:26.057 - 识别结果: 换货。
2026-02-19 14:16:35.711 - INFO - wyoming-funasr-stt-server - Transcribe received
2026-02-19 14:16:35.711 - INFO - wyoming-funasr-stt-server - AudioStart received
2026-02-19 14:16:38.141 - INFO - wyoming-funasr-stt-server - AudioStop received. Processing...  78400 bytes of audio...
2026-02-19 14:16:38.143 - 音频长度: 39200 samples, 2.45 秒
2026-02-19 14:16:38.143 - INFO - wyoming-funasr-stt-server - Audio length: 39200 samples, 2.45 s
rtf_avg: 0.028: 100%|█████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 14.81it/s]
rtf_avg: 1.693: 100%|█████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00,  1.53s/it]
rtf_avg: 0.624, time_speech:  2.450, time_escape: 1.530: 100%|████████████████████████████████████████████| 1/1 [00:01<00:00,  1.53s/it]
2026-02-19 14:16:39.742 - 过滤掉 SenseVoice 可能输出的情感/事件标签
2026-02-19 14:16:39.742 - INFO - wyoming-funasr-stt-server - 识别结果原文：可能输出的情感/事件标签- Result: <|zh|><|NEUTRAL|><|Speech|><|withitn|>关火。
2026-02-19 14:16:39.742 - INFO - wyoming-funasr-stt-server - Result: 关火。
2026-02-19 14:16:39.742 - 识别结果: 关火。

pip3 list

Package                Version
---------------------- --------
aliyun-python-sdk-core 2.16.0
aliyun-python-sdk-kms  2.16.5
antlr4-python3-runtime 4.9.3
audioread              3.1.0
certifi                2026.1.4
cffi                   2.0.0
charset-normalizer     3.4.4
crcmod                 1.7
cryptography           46.0.3
decorator              5.2.1
editdistance           0.8.1
filelock               3.20.3
fsspec                 2026.1.0
funasr                 1.3.0
hydra-core             1.3.2
idna                   3.11
ifaddr                 0.2.0
jaconv                 0.4.1
jamo                   0.4.1
jieba                  0.42.1
Jinja2                 3.1.6
jmespath               0.10.0
joblib                 1.5.3
kaldiio                2.18.1
lazy_loader            0.4
librosa                0.11.0
llvmlite               0.46.0
MarkupSafe             3.0.3
modelscope             1.34.0
mpmath                 1.3.0
msgpack                1.1.2
networkx               3.6.1
numba                  0.63.1
numpy                  1.26.4
omegaconf              2.3.0
oss2                   2.19.1
packaging              26.0
pip                    23.0.1
platformdirs           4.5.1
pooch                  1.8.2
protobuf               6.33.4
pycparser              3.0
pycryptodome           3.23.0
pynndescent            0.6.0
pytorch-wpe            0.0.1
PyYAML                 6.0.3
requests               2.32.5
scikit-learn           1.8.0
scipy                  1.17.0
sentencepiece          0.2.1
setuptools             66.1.1
six                    1.17.0
soundfile              0.13.1
soxr                   1.0.0
sympy                  1.14.0
tensorboardX           2.6.4
threadpoolctl          3.6.0
torch                  2.1.0
torch_complex          0.4.4
torchaudio             2.1.0
tqdm                   4.67.1
typing_extensions      4.15.0
umap-learn             0.5.11
urllib3                2.6.3
wyoming                1.8.0
zeroconf               0.148.0

Strategies to reduce latency

1.Use smaller models

FunASR has paraformer-zh-small or paraformer-zh-medium

2.VAD pre-filtering

Skip silence chunks → speech → Skip silence chunks

Create Wyoming server for Home assistant Part3 - stt - wyoming-funasr arm64 sherpa-onnx

blog.matterxiaomi.com — Wed, 04 Feb 2026 00:53:12 GMT

stt - wyoming-funasr arm64 onnx

Wyoming asr server for Home assistant: Step-by-Step Guide for Developers

Building FunASR with sherpa-onnx on an ARM64 (aarch64) system.

To make an STT server work with Home Assistant, the industry standard is using the Wyoming Protocol.

sherpa-onnx: The "engine" that runs the stt. It manages the model files and utilizes the RPi5’s CPU.

Local Control: You use the Wyoming Protocol Integration bridge to link the stt server to Home Assistant's "Assist" pipeline.

Quick start
Create Python virtual environment
Install sherpa-onnx
- Install
Download Model
- download pre-trained models(SenseVoice)
server.py
- step 1. install numpy

Quick start

# Install sherpa-onnx

pip3 install sherpa-onnx sherpa-onnx-bin

pip3 install wyoming==1.8.0

pip3 install numpy

Create Python virtual environment

mkdir -p /funasr-wyoming-sherpa-onnx

cd /funasr-wyoming-sherpa-onnx

python3 -m venv venv

source venv/bin/activate

Install sherpa-onnx

Install

pip3 install sherpa-onnx sherpa-onnx-bin

Installing collected packages: sherpa-onnx-core, sherpa-onnx-bin, sherpa-onnx
Successfully installed sherpa-onnx-1.12.23 sherpa-onnx-bin-1.12.23 sherpa-onnx-core-1.12.23

https://k2-fsa.github.io/sherpa/onnx/python/install.html#method-1-from-pre-compiled-wheels-cpu-only

# pip3 show sherpa-onnx

output

Name: sherpa_onnx
Version: 1.12.23
Summary: 
Home-page: https://github.com/k2-fsa/sherpa-onnx
Author: The sherpa-onnx development team
Author-email: dpovey@gmail.com
License: Apache licensed, as found in the LICENSE file
Location: /funasr-wyoming-sherpa-onnx/venv/lib/python3.11/site-packages

Download Model

This section describes how to download pre-trained SenseVoice models.

download pre-trained models(SenseVoice)

cd /funasr-wyoming-sherpa-onnx



wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2025-09-09.tar.bz2

tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2025-09-09.tar.bz2

rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2025-09-09.tar.bz2

# 安装 ModelScope
pip install modelscope

# SDK 模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('xiaowangge/sherpa-onnx-sense-voice-small')

output

root@raspberrypi:/funasr-wyoming-sherpa-onnx# tree -L 2
.
├── sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2025-09-09
│   ├── model.int8.onnx
│   ├── README.md
│   ├── test_wavs
│   └── tokens.txt

server.py

step 1. install numpy

pip3 install numpy

output

Installing collected packages: numpy
Successfully installed numpy-2.4.2

pip3 install wyoming==1.8.0

output

  Downloading https://www.piwheels.org/simple/wyoming/wyoming-1.8.0-py3-none-any.whl (39 kB)
Installing collected packages: wyoming
Successfully installed wyoming-1.8.0

cd /funasr-wyoming-sherpa-onnx

source venv/bin/activate

python3 server.py

useful links

how to download pre-trained SenseVoice models.

https://k2-fsa.github.io/sherpa/onnx/sense-voice/index.html

https://k2-fsa.github.io/sherpa/onnx/sense-voice/pretrained.html#sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17

https://k2-fsa.github.io/sherpa/onnx/sense-voice/pretrained.html#sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2025-09-09-chinese-english-japanese-korean-cantonese

How to repeat an action until another (specific) device trigger is received in home assistant

blog.matterxiaomi.com — Sat, 31 Jan 2026 13:53:45 GMT

How to repeat an action until another (specific) device trigger is received in home assistant

To repeat an action in Home Assistant until a specific device trigger is received, use a repeat-until loop combined with a wait_for_trigger action.

until

alias: New automation find my mobile phone - 找手机
description: >-
  - http://localhost:4999/boards/topic/16760/notify - ## 
  http://localhost:4999/boards/topic/16760 - ##
  http://192.168.2.125:8123/config/automation/edit/1747045045725
triggers:
  - trigger: conversation
    command: "[找手机|手机|手机在哪儿|我手机在哪儿]"
    id: id_find_my_phone
  - trigger: state
    entity_id:
      - binary_sensor.sm_g9910_private_interactive
    from:
      - "off"
    to:
      - "on"
    id: id_phone_interactive_turn_on
    enabled: true
conditions: []
actions:


  - repeat:
      until:
        - condition: trigger
          id:
            - id_phone_interactive_turn_on
      sequence:
        - action: script.turn_on
          metadata: {}
          data: {}
          target:
            entity_id: script.find_my_phone
        - delay:
            hours: 0
            minutes: 0
            seconds: 5
            milliseconds: 0
    enabled: false
 

mode: single

way 2.

sequence:
  - variables:
      stopVar: false
  - parallel:
      - repeat:
          until:
            - condition: template
              value_template: "{{ stopVar == true }}"
          sequence:
            - delay:
                hours: 0
                minutes: 0
                seconds: 1
                milliseconds: 0
            - action: input_boolean.toggle
              metadata: {}
              data: {}
              target:
                entity_id: input_boolean.quick_toggle
      - sequence:
          - wait_for_trigger:
              - trigger: state
                entity_id:
                  - button.push
          - variables:
              stopVar: true
alias: Parallel Test
description: ""

How to Install nextcloud on raspberry pi(Without Domain)

blog.matterxiaomi.com — Sun, 14 Dec 2025 16:38:35 GMT

How to Install nextcloud on raspberry pi.

Normal Docker Works Without Domain.Local processing.

Quick Start
Part 1.Installing Nextcloud on a Raspberry Pi with docker
part 2.Use Windows folder share as external storage in Nextcloud
Backups

Quick Start

Raspberry Pi OS Lite (Debian 12 bookworm) Setup

Install Type：docker Nextcloud

Nextcloud + Docker + RPi + Windows SMB, SQLite.

Access Nextcloud via http://{raspberry-pi-ip}:8120

I installed NextCloud Docker Image/Container on my Raspberry. Everything works fine.

Part 1.Installing Nextcloud on a Raspberry Pi with docker

step 1.Run Nextcloud (SQLite, NO DB container,Works via IP)

Note

Uses SQLite by default( No MariaDB / PostgreSQL)

Works via IP(No domain / HTTPS required)

Viewing the Nextcloud configuration (config.php)

# Inside container
docker exec -it nextcloud bash

# Check config:
grep dbtype config/config.php

output

'dbtype' => 'sqlite3',

step 2.First-Time Setup (Web UI)

http://<RPI-IP>:8280

Note

Database:SQLite is selected

part 2.Use Windows folder share as external storage in Nextcloud

Mount shared folder on Windows10 into a folder in your nextcloud drive and use it as data storage.

Docker NextCloud connection via SMB to Windows 10

Create a shared folder on Windows10

Auto-Mount at Boot on rpi5

Mount Windows shared folder into Nextcloud

Backups

Files are located in the NextCloud folder, the data subfolder.

For example, the admin user, is probably in /var/www/html/nextcloud/data/admin/files/

make a backup of /var/www/html/nextcloud/data .

useful links

https://github.com/nextcloud/docker

http://localhost:4999/boards/topic/48316/nextcloud/page/3#82156

Global Variables in HomeAssistant part8 - Creating devices via MQTT templates

blog.matterxiaomi.com — Thu, 04 Dec 2025 21:44:17 GMT

defining MQTT-based sensors , group a bunch of sensors as a device.

mqtt:
  binary_sensor:
  - state_topic: rtl_433/Acurite-986/5197/battery_ok
    json_attributes_topic: rtl_433/Acurite-986/5197
    device_class: battery
    name: "Chest freezer temperature battery state"
    unique_id: chest_freezer_temperature_battery_state
    payload_on: '0'
    payload_off: '1'
    device:
      identifiers: 986-5197
      manufacturer: AcuRite
      name: Chest freezer temperature
      model: 986
      suggested_area: Green bedroom
  sensor:
  - state_topic: rtl_433/Acurite-986/5197/temperature_C
    json_attributes_topic: rtl_433/Acurite-986/5197
    device_class: temperature
    name: "Chest freezer temperature"
    unique_id: chest_freezer_temperature
    unit_of_measurement: "°C"
    value_template: '{{ value | round(0) }}'
    state_class: "measurement"
    device:
      identifiers: 986-5197
      manufacturer: AcuRite
      name: Chest freezer temperature
      model: 986
      suggested_area: Green bedroom

Once defined in YAML, reload the configuration and you'll notice your entities are now groups as a device.

Set attributes only

blog.matterxiaomi.com — Sun, 23 Nov 2025 15:37:31 GMT

How to change an attribute of an entity within an automation（）

a python_script can be used to set the state and attributes of any entity programmatically.

You can also add or modify attributes associated with the entity in the "Attributes" field

Enable python script integration in configuration.yaml

Enable python script integration in configuration.yaml

Add the component to your configuration.yaml file.

python_script:

Python script

# /config/python_scripts/update_entity_attributes.py
# author：matterxiaomi
# blog：

entity_id = data.get('entity_id')
attributes_to_update = data.get('attributes')

if entity_id is None:
    logger.error("entity_id is required for updating attributes.")
elif attributes_to_update is None:
    logger.warning("No attributes provided to update for entity: %s", entity_id)
else:
    # Get current state and attributes
    current_state_object = hass.states.get(entity_id)

    if current_state_object:
        current_state = current_state_object.state
        current_attributes = dict(current_state_object.attributes) # Create a mutable copy

        # Update only the specified attributes
        current_attributes.update(attributes_to_update)

        # Set the state back with updated attributes, preserving the original state
        hass.states.set(entity_id, current_state, current_attributes)
        logger.info("Attributes updated for entity %s", entity_id)
    else:
        logger.warning("Entity %s not found. Cannot update attributes.", entity_id)

Explanation:

1.Set attributes only

2.The existing state will be preserved unless explicitly provided in the state parameter.

Developer tools

action: python_script.update_entity_attributes
data: 
  entity_id: sensor.my_map
  attributes:
    friendly_name: new_value

use this script in an automation or another script:

- service: python_script.update_entity_attributes
  data:
    entity_id: sensor.my_entity
    attributes:
      new_attribute_1: value_1
      another_attribute: new_value

Explanation:

To update only attributes of an entity in Home Assistant using a Python script, you can leverage the hass.states.set() function, passing in the entity_id and the new_attributes you want to set. The existing state will be preserved unless explicitly provided in the state parameter.

homeassistant.update_entity

script:
  update_travel_time:
    alias: "Manually update travel time sensor"
    sequence:
      - service: homeassistant.update_entity
        target:
          entity_id: sensor.google_travel_time_home_work # Replace with your sensor's entity ID

mkdir config/python_scripts
cat <<'EOF' > config/python_scripts/set_state.py
entity = data['entity_id']
hass.states.set(entity, data['state'])
EOF

Set State via Button

square: true
type: grid
cards:
  - show_name: true
    show_icon: true
    name: sleep
    type: button
    tap_action:
      action: call-service
      service: python_script.set_state
      data:
        entity_id: sensor.baby
        state: sleep
      target: {}
    icon: mdi:baby-face-outline

useful links

Python script for Home Assistant to handle state and attributes of existing sensors and entities.

https://github.com/pmazz/ps_hassio_entities

Ha Network Storage

blog.matterxiaomi.com — Sun, 23 Nov 2025 15:16:36 GMT

The External Storage Support application enables you to mount external storage services and devices as secondary HA storage devices.

Client
- rpi5 -> smb folder share on windows 10

storage server and protocol

storage client and protocol

# protocol：
# - SMB: smb://192.168.1.100/PhoneShare
# - FTP: ftp://192.168.1.100:21
# - WebDAV: http://192.168.1.100:8080/
# - SFTP: sftp://192.168.1.100:22
# - NFS: smount -t nfs 192.168.1.10:/home/share /mnt

				ha on rpi5		Scene
Server				Client
OS Type	Inside OS	sharing protocol		Inside OS	Inside HA
local storage server
windows 10	Install IIS WebDAV Server	Web server	webdav Storage		webdav integration

windows 10	Install NFS server		NFS Storage	Insall nfs-common
windows 10	SMB Storage		SMB Storage	Install smbclient cifs-utils		ha Use windows share folder as data storage

		FTP/FTPS
linux	SFTP Server	SFTP over SSH(SSH File transfer protocol)	SFTP Storage
linux	Install NFS server			`sudo mount -t nfs 192.168.1.10:/home/share /mnt`

Usb external storage						buy
remote cloud storage		Amazon S3，Dropbox, Google Drive, Amazon

Client

Client on rpi5（OS）

rpi5 -> smb folder share on windows 10

SMB - use cifs

sudo mount -t cifs -o username=your_username,password=your_password //server_ip/share_name /mnt/smb_share

SMB - not use cifs

smbclient -L //192.168.2.21 -U test2

output

	Sharename       Type      Comment
	---------       ----      -------

	nextcloud       Disk      
	shareforgiteagitserver Disk

smbclient //192.168.2.21/nextcloud -U username

output

smb: \> l
  .                                   D        0  Mon Dec 15 03:50:39 2025
  ..                                  D        0  Mon Dec 15 03:50:39 2025
  data                                D        0  Mon Dec 15 03:55:30 2025

		732562175 blocks of size 4096. 91858085 blocks available
smb: \>

blog.matterxiaomi.com

Matter Bridge in Home Assistant Part2 - Install MatterBridge Connect to Home assistant

Table of Contents

Quick start

Install and configure

How to Use

Ecovacs in Home Assistant Part6 - Robot Vacuum Control MCP Server

ModelScope vs Hugging Face vs k2-fsa.github.io vs Kaldi vs Sherpa

Table of Contents

Hugging Face（Global AI model platform）

ModelScope(The Alibaba/Chinese AI Industrial model platform)

specialized tools focused on speech recognition (ASR) and synthesis (TTS)

k2-fsa (Next-Gen Kaldi)

download Silero VAD ONNX model

Sherpa(The Real-Time Speech Deployment Tool)

Create Wyoming server for Home assistant Part2 - stt - wyoming-funasr arm64

Table of Contents

Step 1.Development Environment Setup

Create Python virtual environment

Step 2. Install

Install torch via PyPI

Install FunASR 1.3.0 via PyPI

Verify installation

Step3.Download and test a model (example: paraformer-zh)

SenseVoice - Speech Recognition (Non-streaming)

Step 4.FunASR + Wyoming STT full server

Install wyoming

Strategies to reduce latency

Create Wyoming server for Home assistant Part3 - stt - wyoming-funasr arm64 sherpa-onnx

Table of Contents

Quick start

Create Python virtual environment

Install sherpa-onnx

Install

Download Model

download pre-trained models(SenseVoice)

server.py

step 1. install numpy

How to repeat an action until another (specific) device trigger is received in home assistant

How to Install nextcloud on raspberry pi(Without Domain)

Table of Contents

Quick Start

Part 1.Installing Nextcloud on a Raspberry Pi with docker

part 2.Use Windows folder share as external storage in Nextcloud

Backups

Global Variables in HomeAssistant part8 - Creating devices via MQTT templates

Set attributes only

Table of Contents

Enable python script integration in configuration.yaml

Ha Network Storage

Table of Contents

Client

rpi5 -> smb folder share on windows 10