M5 Atom Echo - No text recognized (stt-no-text-recognized)

ESPHome
1

I’ve bought the M5 Atom Echo, got through all the steps to make it work, and it does fairly well, when I’m using a fully local stack (ollama, piper, whisper all running in a docker container).

My problem comes when I try to use it with Home Assistant Cloud.
The device is online, it responds to it’s wake-word, but right after, it fails No text recognized (stt-no-text-recognized).

The assistant works independently - if I talk to it through the web interface, with voice - it performs the action as expected.

Here is the log of the failed attempt speaking to the M5 Atom Echo:

stage: done
run:
  pipeline: 01kc77shx42fje3qksbf49mwbg
  language: en
  conversation_id: 01KDK8DDQPY3931RWK9VT3S59P
  satellite_id: assist_satellite.m5stack_atom_echo_bcf658_assist_satellite
  tts_output:
    token: 2BWBJWOH6fLcnpD2bi2pVg.wav
    url: /api/tts_proxy/2BWBJWOH6fLcnpD2bi2pVg.wav
    mime_type: audio/x-wav
    stream_response: false
events:
  - type: run-start
    data:
      pipeline: 01kc77shx42fje3qksbf49mwbg
      language: en
      conversation_id: 01KDK8DDQPY3931RWK9VT3S59P
      satellite_id: assist_satellite.m5stack_atom_echo_bcf658_assist_satellite
      tts_output:
        token: 2BWBJWOH6fLcnpD2bi2pVg.wav
        url: /api/tts_proxy/2BWBJWOH6fLcnpD2bi2pVg.wav
        mime_type: audio/x-wav
        stream_response: false
    timestamp: "2025-12-28T19:55:39.071165+00:00"
  - type: wake_word-start
    data:
      entity_id: wake_word.openwakeword
      metadata:
        format: wav
        codec: pcm
        bit_rate: 16
        sample_rate: 16000
        channel: 1
      timeout: 0
    timestamp: "2025-12-28T19:55:39.071446+00:00"
  - type: wake_word-end
    data:
      wake_word_output:
        wake_word_id: lumi
        wake_word_phrase: Lumi
        timestamp: 327860
    timestamp: "2025-12-28T20:01:06.976699+00:00"
  - type: stt-start
    data:
      engine: stt.home_assistant_cloud
      metadata:
        language: en-US
        format: wav
        codec: pcm
        bit_rate: 16
        sample_rate: 16000
        channel: 1
    timestamp: "2025-12-28T20:01:06.976886+00:00"
  - type: stt-vad-start
    data:
      timestamp: 329180
    timestamp: "2025-12-28T20:01:08.288219+00:00"
  - type: stt-vad-end
    data:
      timestamp: 330360
    timestamp: "2025-12-28T20:01:09.463564+00:00"
  - type: error
    data:
      code: stt-no-text-recognized
      message: No text recognized
    timestamp: "2025-12-28T20:01:09.499367+00:00"
  - type: run-end
    data: null
    timestamp: "2025-12-28T20:01:09.499729+00:00"
started: 2025-12-28T19:55:39.071Z
wake_word:
  entity_id: wake_word.openwakeword
  metadata:
    format: wav
    codec: pcm
    bit_rate: 16
    sample_rate: 16000
    channel: 1
  timeout: 0
  done: true
  wake_word_output:
    wake_word_id: lumi
    wake_word_phrase: Lumi
    timestamp: 327860
stt:
  engine: stt.home_assistant_cloud
  metadata:
    language: en-US
    format: wav
    codec: pcm
    bit_rate: 16
    sample_rate: 16000
    channel: 1
  done: false
finished: 2025-12-28T20:01:09.499Z
error:
  code: stt-no-text-recognized
  message: No text recognized

Does anybody have a clue what the problem might be based on this description?

2

HA cloud uses speech to phrase for STT provider

“No text recognized” means what you said do not match one of the predefined sentences it matches against

Try simple “what time is it” command.

Try speaking louder or try verifying the sentence you are steaming matches the defined phrases