[cursor]

Regenerated examples after elevenlabs/skills#146 (Update skills from changelog 2026-10-26) merged.

pnpm generate summary

0 generated · 19 verified · 1 updated · 0 failed
Total time: 1h 26m 25s

Example	Result
agents/expo/quickstart	Verified, no changes
agents/nextjs/guardrails	Verified, no changes
agents/nextjs/openai-realtime	Verified, no changes
agents/nextjs/quickstart	Verified, no changes
dubbing/nextjs/quickstart	Verified, no changes
music/nextjs/quickstart	Verified, no changes
music/python/quickstart	Verified, no changes
music/typescript/quickstart	Verified, no changes
sound-effects/nextjs/quickstart	Verified, no changes
sound-effects/python/quickstart	Verified, no changes
sound-effects/typescript/quickstart	Verified, no changes
speech-engine/nextjs/quickstart	Verified, no changes
speech-to-text/nextjs/realtime	Verified, no changes
speech-to-text/python/quickstart	Updated
speech-to-text/typescript/quickstart	Verified, no changes
text-to-speech/nextjs/quickstart	Verified, no changes
text-to-speech/python/quickstart	Verified, no changes
text-to-speech/typescript/quickstart	Verified, no changes
voice-changer/nextjs/quickstart	Verified, no changes
voice-isolator/nextjs/quickstart	Verified, no changes

Changed directories

 speech-to-text/python/quickstart/example/main.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

Are the changes meaningful?

This is a small but real fix. It is not driven by skills PR #146, whose changes (agent service_tier, start_procedure_ids, widget mobile-fullscreen, TTS clone_strength, Voice Changer REST path wording) are docs-only and don't touch anything these examples use.

client.speech_to_text.convert() returns a union of SpeechToTextChunkResponseModel, MultichannelSpeechToTextResponseModel and SpeechToTextWebhookResponseModel (elevenlabs 2.75.0). The old code read result.text directly, and pyright reports 2 errors because the multichannel and webhook variants have no text attribute. The update narrows with isinstance(result, SpeechToTextChunkResponseModel) and exits with a readable error otherwise. Runtime behaviour for this single-channel, non-webhook call is unchanged.

Checked locally: pyright main.py (0 errors, down from 2 before the change) and ruff check main.py (passes).

Note on the run

The dependency-install step for music/typescript/quickstart hung for about 40 minutes. The example pins packageManager: pnpm@9.15.4, and corepack (COREPACK_ENABLE_DOWNLOAD_PROMPT=1) waited on an interactive download prompt whose text was hidden by the script's output capture. That step isn't covered by the 600s timeout. I answered the prompt manually and the run continued. A follow-up could set COREPACK_ENABLE_DOWNLOAD_PROMPT=0 or redirect stdin from /dev/null in install_example_dependencies.

  

[tadaspetra]

update seems minor