Capability coverage one console · local first
HTTP ASR / TTSOpenAI-compatible transcription and speech endpoints.
Streaming ASRMicrophone or file input, pure stream and VAD segmentation.
Streaming TTSTrue chunk streaming, sentence segmentation, latency and RTF.
Observer bridgeOptional ffmpeg source, video, RSS/CPU and soak-test alerts.
Model selection registry · download · hot load
Connect to load the model registry.
HTTP transcription POST /v1/audio/transcriptions
HTTP speech POST /v1/audio/speech
WS streaming ASR /asr/stream
Audio
0.0s
16 kHz mono
Level
-120 dB
peak -120 dB
Messages
0
0 finals
ASR idle
0s
since last event
WS VAD + ASR /asr/vad
WS TTS /tts/stream · /tts/segmented
First audio
-
request to first chunk
Chunks
0
0 segments
Audio
0.00s
received PCM
Wall / RTF
-
RTF -
Buffer ahead
-
positive is healthy
This bridge keeps ffmpeg, video preview, process RSS/CPU, idle and RSS-slope checks in the existing observer backend. Start it separately with
python3 examples/server/asr-observer/observer.py.Source audio
0.0s
0.00 MB
Level
-120 dB
peak -120 dB
rs-server
n/a
cpu n/a
ffmpeg
n/a
cpu n/a
Idle
0s
source 0s
RSS slope
n/a
ffmpeg n/a
No video frames
No observer alerts