mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-07-09 13:21:42 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

fix: gate interrupt STT transcript echoes

Browse source
This commit is contained in:
devatnull 2026-06-28 16:19:25 +03:00 • committed by Teknium
parent bfc5262725
commit 406eb719c3
2 changed files with 23 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

12
gateway/run.py
View file

@ -18613,7 +18613,7 @@ class GatewayRunner(GatewayAuthorizationMixin, GatewayKanbanWatchersMixin, Gatew
# real transcript instead of an empty string
# (or file-path placeholder). Matches the UX
# of fresh voice messages including the
# 🎙️ echo back to the user.
# optional 🎙️ echo back to the user.
_media_urls = getattr(_peek_event, "media_urls", None) or []
_media_types = getattr(_peek_event, "media_types", None) or []
_audio_paths = []
@ -18631,7 +18631,7 @@ class GatewayRunner(GatewayAuthorizationMixin, GatewayKanbanWatchersMixin, Gatew
pending_text, _audio_paths,
)
pending_text = _enriched
if _transcripts:
if _transcripts and self._should_echo_stt_transcripts():
_echo_meta = {"thread_id": source.thread_id} if source.thread_id else None
for _tx in _transcripts:
try:
@ -19030,9 +19030,9 @@ class GatewayRunner(GatewayAuthorizationMixin, GatewayKanbanWatchersMixin, Gatew
# Transcribe audio media on the dequeued event BEFORE it is
# handed back as the next user turn, so queued/interrupting
# voice messages drain with the real transcript instead of
# a file-path placeholder. Echo each transcript back to the
# user (same 🎙️ format as fresh voice messages) so voice
# interrupts feel identical to text interrupts.
# a file-path placeholder. When configured, echo each
# transcript back to the user in the same 🎙️ format as
# fresh voice messages.
_pending_text = pending_event.text or ""
_media_urls = getattr(pending_event, "media_urls", None) or []
_media_types = getattr(pending_event, "media_types", None) or []
@ -19051,7 +19051,7 @@ class GatewayRunner(GatewayAuthorizationMixin, GatewayKanbanWatchersMixin, Gatew
_pending_text, _audio_paths,
)
pending = _enriched or None
if _transcripts:
if _transcripts and self._should_echo_stt_transcripts():
_echo_meta = {"thread_id": source.thread_id} if source.thread_id else None
for _tx in _transcripts:
try:

17
tests/gateway/test_stt_transcript_echo_config.py
View file

@ -1,3 +1,4 @@
from pathlib import Path
from types import SimpleNamespace
from gateway.config import GatewayConfig
@ -39,3 +40,19 @@ def test_gateway_runner_uses_stt_echo_transcripts_flag():
runner.config = SimpleNamespace()
assert runner._should_echo_stt_transcripts() is True
def test_all_gateway_transcript_echo_sends_are_gated():
source = Path(__file__).resolve().parents[2] / "gateway" / "run.py"
lines = source.read_text().splitlines()
echo_send_lines = [
index
for index, line in enumerate(lines)
if "f'🎙️" in line or 'f"🎙️' in line
]
assert echo_send_lines
for index in echo_send_lines:
context = "\n".join(lines[max(0, index - 12): index + 1])
assert "_should_echo_stt_transcripts()" in context