Whisper fails wyoming_faster_whisper.sherpa_handler:Downloading https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8.tar.bz2

[SamuelCarson]

Describe the issue you are experiencing

Whisper fails repeatedly while downloading.
wyoming_faster_whisper.sherpa_handler:Downloading https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8.tar.bz2

What type of installation are you running?

Home Assistant OS

Which operating system are you running on?

Home Assistant Operating System

Which app are you reporting an issue with?

Whisper

What is the version of the app?

3.5.1

Steps to reproduce the issue

1.Install Whisper app
2.Start App
3.App fails
...

System Health information

System Information

version	core-2026.7.4
installation_type	Home Assistant OS
dev	false
hassio	true
docker	true
container_arch	aarch64
user	root
virtualenv	false
python_version	3.14.6
os_name	Linux
os_version	6.18.34-haos-raspi
arch	aarch64
timezone	America/New_York
config_dir	/config

Home Assistant Cloud

logged_in	false
can_reach_cert_server	ok
can_reach_cloud_auth	ok
can_reach_cloud	ok

HACS

error	failed to load: unknown

Home Assistant Supervisor

host_os	Home Assistant OS 18.2
update_channel	stable
supervisor_version	supervisor-2026.07.5
agent_version	1.10.0
docker_version	29.6.2
disk_total	468.7 GB
disk_used	73.9 GB
nameservers	1.1.1.1, 192.168.1.1
healthy	true
supported	true
host_connectivity	true
supervisor_connectivity	true
ntp_synchronized	true
virtualization
board	rpi5-64
disk_life_time	3 %
supervisor_api	ok
version_api	ok
installed_addons	File editor (6.1.0), Node-RED (22.0.1), WireGuard (0.13.0), InfluxDB (5.0.2), AdGuard Home (6.2.0), Terminal & SSH (10.3.0), TasmoBackup (1.08.00), Grafana (12.1.0), NGINX Home Assistant SSL proxy (4.5.1), MariaDB (3.0.1), Samba share (12.10.0), Network UPS Tools (0.18.0), Mosquitto broker (6.5.2), TasmoAdmin (0.33.0), ESPHome Device Builder (2026.7.3), Matter Server (9.1.1), OpenThread Border Router (3.1.0), Advanced SSH & Web Terminal (24.0.1), Assist Satellite (1.1.15), Whisper (3.5.1), Piper (2.3.2)

Dashboards

dashboards	4
resources	7
views	1
mode	storage

Network Configuration

adapters	lo (disabled), end0 (enabled, default, auto), docker0 (disabled), hassio (disabled), veth3cb6f55 (disabled), veth5483f88 (disabled), veth6b0b06a (disabled), vethb393a83 (disabled), veth540ac59 (disabled), veth6a4d0b9 (disabled), veth285fc70 (disabled), veth3af443b (disabled), veth3e95d9a (disabled), vethf9c457a (disabled), veth6ff165e (disabled), veth52fd14b (disabled), vethd61f0b4 (disabled), veth7625c6d (disabled), veth5cdc4f2 (disabled), vetha5539a1 (disabled), veth5a55a5a (disabled), vetha3da32a (disabled)
ipv4_addresses	lo (127.0.0.1/8), end0 (192.168.1.254/24), docker0 (172.30.232.1/23), hassio (172.30.32.1/23), veth3cb6f55 (), veth5483f88 (), veth6b0b06a (), vethb393a83 (), veth540ac59 (), veth6a4d0b9 (), veth285fc70 (), veth3af443b (), veth3e95d9a (), vethf9c457a (), veth6ff165e (), veth52fd14b (), vethd61f0b4 (), veth7625c6d (), veth5cdc4f2 (), vetha5539a1 (), veth5a55a5a (), vetha3da32a ()
ipv6_addresses	lo (::1/128), end0 (fe80::ab3d:644d:a732:3770/64), docker0 (fdab:54e8:e942::1/64, fe80::58e9:31ff:fed2:e6e0/64), hassio (fd0c:ac1e:2100::1/48, fe80::24bc:50ff:fe2b:55df/64), veth3cb6f55 (fe80::9c52:37ff:fe55:756c/64), veth5483f88 (fe80::1c38:f5ff:fe41:c239/64), veth6b0b06a (fe80::807f:2aff:fe80:2dae/64), vethb393a83 (fe80::f4bd:67ff:fedf:58fa/64), veth540ac59 (fe80::6883:acff:fef9:1ec9/64), veth6a4d0b9 (fe80::c49d:1bff:fef3:381/64), veth285fc70 (fe80::6846:d0ff:fe0b:7b9b/64), veth3af443b (fe80::7c8b:c8ff:fe24:6fde/64), veth3e95d9a (fe80::683f:d6ff:feec:c0ce/64), vethf9c457a (fe80::4cfb:3ff:fe27:ebf5/64), veth6ff165e (fe80::f443:38ff:fe12:fcb6/64), veth52fd14b (fe80::e4a7:1bff:fe4a:835d/64), vethd61f0b4 (fe80::a0b3:22ff:fe6a:e3fc/64), veth7625c6d (fe80::3881:fdff:febb:dc8e/64), veth5cdc4f2 (fe80::804f:4fff:fe36:9087/64), vetha5539a1 (fe80::bc0b:c2ff:feec:a4b/64), veth5a55a5a (fe80::c0f8:6ff:fe7e:54dc/64), vetha3da32a (fe80::fc37:c7ff:fecf:d31a/64)
announce_addresses	192.168.1.254, fe80::ab3d:644d:a732:3770

Recorder

oldest_recorder_run	August 2, 2026 at 8:21 PM
current_recorder_run	August 4, 2026 at 5:10 PM
estimated_db_size	976.22 MiB
database_engine	mysql
database_version	11.4.10

Anything in the Supervisor logs that might be useful for us?

no

Anything in the app logs that might be useful for us?

Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/__main__.py", line 288, in <module>
    run()
    ~~~^^
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/__main__.py", line 283, in run
    asyncio.run(main())
    ~~~~~~~~~~~^^^^^^^^
  File "/usr/lib/python3.13/asyncio/runners.py", line 195, in run
    return runner.run(main)
           ~~~~~~~~~~^^^^^^
  File "/usr/lib/python3.13/asyncio/runners.py", line 118, in run
    return self._loop.run_until_complete(task)
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^
  File "/usr/lib/python3.13/asyncio/base_events.py", line 725, in run_until_complete
    return future.result()
           ~~~~~~~~~~~~~^^
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/__main__.py", line 252, in main
    await loader.load_transcriber()
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/models.py", line 158, in load_transcriber
    transcriber = SherpaTranscriber(
        model, self.download_dir, cpu_threads=self.cpu_threads
    )
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/sherpa_handler.py", line 89, in __init__
    self.recognizer = so.OfflineRecognizer.from_transducer(
                      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
        num_threads=cpu_threads,
        ^^^^^^^^^^^^^^^^^^^^^^^^
    ...<5 lines>...
        model_type="nemo_transducer",
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    )
    ^
  File "/usr/src/.venv/lib/python3.13/site-packages/sherpa_onnx/offline_recognizer.py", line 262, in from_transducer
    self.recognizer = _Recognizer(recognizer_config)
                      ~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^
RuntimeError: Load model from /data/sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8/encoder.int8.onnx failed:Protobuf parsing failed.
[09:47:09] INFO: Service exited with code 1 (by signal 0)
Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/__main__.py", line 288, in <module>
    run()
    ~~~^^
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/__main__.py", line 283, in run
    asyncio.run(main())
    ~~~~~~~~~~~^^^^^^^^
  File "/usr/lib/python3.13/asyncio/runners.py", line 195, in run
    return runner.run(main)
           ~~~~~~~~~~^^^^^^
  File "/usr/lib/python3.13/asyncio/runners.py", line 118, in run
    return self._loop.run_until_complete(task)
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^
  File "/usr/lib/python3.13/asyncio/base_events.py", line 725, in run_until_complete
    return future.result()
           ~~~~~~~~~~~~~^^
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/__main__.py", line 252, in main
    await loader.load_transcriber()
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/models.py", line 158, in load_transcriber
    transcriber = SherpaTranscriber(
        model, self.download_dir, cpu_threads=self.cpu_threads
    )
  File "/usr/src/.venv/lib/python3.13/site-packages/wyoming_faster_whisper/sherpa_handler.py", line 89, in __init__
    self.recognizer = so.OfflineRecognizer.from_transducer(
                      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
        num_threads=cpu_threads,
        ^^^^^^^^^^^^^^^^^^^^^^^^
    ...<5 lines>...
        model_type="nemo_transducer",
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    )
    ^
  File "/usr/src/.venv/lib/python3.13/site-packages/sherpa_onnx/offline_recognizer.py", line 262, in from_transducer
    self.recognizer = _Recognizer(recognizer_config)
                      ~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^
RuntimeError: Load model from /data/sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8/encoder.int8.onnx failed:Protobuf parsing failed.
[09:47:16] INFO: Service exited with code 1 (by signal 0)


Then repeated ad nauseum.

Additional information

No response

[alexp999]

I've got HAOS running as a VM on an x86 machine. I encountered this too and it completely locks up the VM, the CPU usage spikes and HA becomes unresponsive, I have to force reboot the machine.
I thought I got round it by setting the library to faster sherpa, which worked for about 5 minutes, before the app spiked CPU and locked up the whole HA and HAOS VM again.
I'm using speech to phrase until I can get this working. I wonder if it's the rebase to trixie for the container? Was it working for you before 3.5.1? Ive only just installed it so no prior experience, but wondering if I can somehow install 3.5.0.

[alexp999]

This turned out to be a heat/cpu thing for me.

The system was crashing due to the immense CPU load in the hot weather which a very dried out thermal paste it turns out. Fresh paste and better CPU cooler, I've not had any issues.