# NousResearch/hermes-agent · Local Whisper STT stalls when HuggingFace Hub unreachable

## Open now

- 0 issues, 1 PR open on GitHub.
- PR #111085: fix(stt): prefer cached local whisper models

A bug causes local speech-to-text using Whisper to silently stall or fail when the HuggingFace Hub is inaccessible, because the code performs a network round-trip even for cached models. Several fix attempts address loading cached models offline and documenting environment variable workarounds for network-restricted hosts.

## TL;DR

- 6 member(s): 0 open issue(s), 1 closed issue(s), 1 open PR(s), 1 merged PR(s), 3 PR(s) closed without merging.
- A fix was merged to main in PR #111359. Verify it addresses your case.
- **Recommended:** read PR #111359 (and any competing PRs below) before filing. Confirm the merged change covers your symptom.

## Maintainer resolution verdict

- Internal-tracker verdict over this cluster's issue members: `resolved` · deciding PR(s): `pr-111359`. This is distinct from the fix status above. Weigh it yourself.

## How this cluster was grouped

- Link evidence (`STRUCTURAL_SPANNING`): every member is connected by structural evidence: shared changed files, or a closing/superseded reference.
- Pairwise link density `0.33`: the share of member pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. `1.00` means every pair is linked.

## Closed / duplicate issues

- #111072: \[Bug\]: Local STT silently stalls ~173 s (or hard-fails) on hosts where huggingface.co is unreachable — HF_ENDPOINT / HF_HUB_DISABLE_XET workarounds undocumented

## Prior & competing PRs

- **#111359: fix(stt): local faster-whisper loads a cached model without a Hub round-trip and names the HF mirror levers (#111072, salvage #111085)**. merged to main · fix tier `verified`
  - maintainer direction: category `merged`
  - files: `tests/tools/test_transcription_tools.py`, `tools/transcription_local.py`, `tools/transcription_tools.py`, `website/docs/user-guide/features/tts.md`, `website/docs/user-guide/features/voice-mode.md`
- **#110883: fix(stt): make the local-whisper CUDA → CPU rescue actually fire**. closed without merge
  - maintainer direction: category `merged_via`; implication `superseded`; refs `a1cbed3592`, `#112167`
- **#111085: fix(stt): prefer cached local whisper models**. open
  - maintainer direction: category `partial_fix`; implication `needs_llm`; refs `#111359`, `#111086`
  - files: `tests/tools/test_transcription_tools.py`, `tools/transcription_local.py`, `tools/transcription_tools.py`, `website/docs/user-guide/features/tts.md`, `website/docs/user-guide/features/voice-mode.md`
- **#111086: fix(stt): load local whisper cache-first and name HF mirror escape hatches**. closed without merge
  - maintainer direction: category `duplicate_of`; implication `duplicate`; refs `8d3bee50dcfdd4679e2060de0c7f6b9be69c7f02`, `440076a316ea0ddd0bdc83233a11d960a4684a60`, `#111085`, `#111072`, `#110883`, `#103848`, `#105336`, `#110679`, `#75554`
- **#111581: fix(stt): load a cached whisper model offline instead of stalling on a revision lookup**. closed without merge
  - maintainer direction: category `duplicate_of`; implication `duplicate`; refs `#111359`, `#111072`

## Fix status

- Merged to main in PR #111359 (cited above). No release data is tracked. Verify downstream.
- Fix last verified (merged) `2026-09-15`.
- Subsystem: `agent`