fix(gateway): inbound voice classification/routing for Feishu, DingTalk, LINE, QQ, Photon, WhatsApp, Weixin - #73515

Merged

teknium1 merged 22 commits into

mainfrom

fix/voice-platform-inbound

Jul 28, 2026

Commits on Jul 28, 2026

fix(feishu): classify native voice messages as VOICE for auto-transcription

wuli666
authored and
teknium1
committed
148e52e
fix(feishu): voice-note duration on upload, turn-scoped TTS dedup, audio path dedup

seamusmore
authored and
teknium1
committed
8361b77
fix(dingtalk-platform): extract ASR recognition and file message text from incoming messages

raywu
authored and
teknium1
committed
e92753d
fix(dingtalk): add EXT_MAP constant, PHOTO classification for images, card/interactiveCard message handling

raywu
authored and
teknium1
committed
38e4efe
fix(dingtalk): don't let richText re-derivation clobber VOICE classification
teknium1
committed
e55f40c
fix(line): normalize inbound media types and cache routing

kronexoi
authored and
teknium1
committed
32b54a8
fix(qqbot): stop routing file uploads through STT pipeline

Zioywishing
authored and
teknium1
committed
a950dcd
test(qqbot): update voice detection tests for content_type-only logic

Zioywishing
authored and
teknium1
committed
2239791
fix(qqbot): keep extension fallback for voice detection, skip only for explicit file uploads

Zioywishing
authored and
teknium1
committed
39b0a9b
fix(qqbot): skip voice detection for file uploads (content_type='file')

Zioywishing
authored and
teknium1
committed
f015150
docs(qqbot): clarify comment on file upload guard

Zioywishing
authored and
teknium1
committed
242d968
fix(qqbot): always clean up temp stt wav

Zhekinmaksim
authored and
teknium1
committed
de2c0c3
fix(photon): map audio/x-caf to .caf extension

Florian Valade
authored and
teknium1
committed
9c5c626
fix(photon): promote .caf attachments to MessageType.VOICE

Florian Valade
authored and
teknium1
committed
14a2646
fix(photon): handle U+FFFC placeholder with deferred wait

Florian Valade
authored and
teknium1
committed
cb90e45
feat(stt): add CAF format support with WAV conversion for cloud providers

Florian Valade
authored and
teknium1
committed
69a9fac
fix(photon): address review — U+FFFC before _record_last_inbound, MIME-based CAF promotion, add tests

Florian Valade
authored and
teknium1
committed
ce9ff2c
fix(whatsapp): preserve voice notes when STT fails

VIVAAN-DHAWAN
authored and
teknium1
committed
95fd0d7
fix(gateway/weixin): route voice messages through Hermes STT instead of trusting Tencent Cloud's text

Kewe63
authored and
teknium1
committed
5303324
test(gateway/weixin): add integration handoff regression for #27300 voice routing

Kewe63
authored and
teknium1
committed
8abb0cc
fix(weixin): preserve voice transcript origin

zgzczzw
authored and
teknium1
committed
312bf30
chore: contributor email mappings for voice-platform-inbound salvage
teknium1
committed
f85b565

Navigation Menu

fix(gateway): inbound voice classification/routing for Feishu, DingTalk, LINE, QQ, Photon, WhatsApp, Weixin - #73515

fix(gateway): inbound voice classification/routing for Feishu, DingTalk, LINE, QQ, Photon, WhatsApp, Weixin#73515
teknium1 merged 22 commits into
mainfrom
fix/voice-platform-inbound

Commits on Jul 28, 2026

fix(feishu): classify native voice messages as VOICE for auto-transcription

fix(feishu): voice-note duration on upload, turn-scoped TTS dedup, audio path dedup

fix(dingtalk-platform): extract ASR recognition and file message text from incoming messages

fix(dingtalk): add EXT_MAP constant, PHOTO classification for images, card/interactiveCard message handling

fix(dingtalk): don't let richText re-derivation clobber VOICE classification

fix(line): normalize inbound media types and cache routing

fix(qqbot): stop routing file uploads through STT pipeline

test(qqbot): update voice detection tests for content_type-only logic

fix(qqbot): keep extension fallback for voice detection, skip only for explicit file uploads

fix(qqbot): skip voice detection for file uploads (content_type='file')

docs(qqbot): clarify comment on file upload guard

fix(qqbot): always clean up temp stt wav

fix(photon): map audio/x-caf to .caf extension

fix(photon): promote .caf attachments to MessageType.VOICE

fix(photon): handle U+FFFC placeholder with deferred wait

feat(stt): add CAF format support with WAV conversion for cloud providers

fix(photon): address review — U+FFFC before _record_last_inbound, MIME-based CAF promotion, add tests

fix(whatsapp): preserve voice notes when STT fails

fix(gateway/weixin): route voice messages through Hermes STT instead of trusting Tencent Cloud's text

test(gateway/weixin): add integration handoff regression for #27300 voice routing

fix(weixin): preserve voice transcript origin

chore: contributor email mappings for voice-platform-inbound salvage