Files
LocalAI/scripts/realtime-conformance.sh
Richard Palethorpe 5d0c43ec6e feat(realtime): Semantic VAD EOU token (#10444)
* feat(realtime): EOU-driven semantic_vad turn detection

Add a `semantic_vad` turn-detection mode to the realtime API that feeds
the transcription model live and decides "the user finished speaking"
from the `<EOU>` end-of-utterance token rather than from silence alone.
When EOU fires the turn commits immediately (~0.3s); otherwise it falls
back to an eagerness-scaled silence threshold (low/med/high = 8/4/2s).

Plumbing, bottom to top:

- proto: `AudioTranscriptionLive` bidirectional RPC (config-first oneof,
  mono float PCM @16k, ready-ack / Unimplemented degrade signal) plus
  `TranscriptResult.eou` for the unary retranscribe gate.
- pkg/grpc: client/server/base/embed scaffolding for the bidi stream,
  modeled on AudioTransformStream; release stream conns on terminal Recv.
- parakeet-cpp: live transcription RPC with per-C-call engine locking
  (one live stream per turn, finalize+free at commit); bump parakeet.cpp
  to ABI v5 — incremental StreamingMel (no more quadratic per-feed mel
  recompute that delayed EOU on long turns) and the <EOU>/<EOB> split;
  strip the literal <EOU>/<EOB> from offline text and set Eou.
- core/backend: LiveTranscriptionSession wrapper + pipeline
  `turn_detection:` config block (type/eagerness/retranscribe).
- realtime: semantic_vad integration — live input captions streamed as
  transcription deltas while the user speaks, EOU-immediate commit with
  eagerness fallback, optional retranscribe gate (batch re-decode must
  also end in <EOU> to confirm), clause synthesis off the LLM token
  callback, and per-turn live-transcription / model_load telemetry.
- UI: show the realtime pipeline components as a vertical list.

Docs and tests included; opt-in via the pipeline YAML or per-session
`session.update`. Non-streaming STT backends degrade to silence-only.

Assisted-by: Claude Code:claude-opus-4-8 [Read] [Edit] [Write] [Bash]
Assisted-by: Claude Code:claude-fable-5 [Read] [Edit] [Bash]
Signed-off-by: Richard Palethorpe <io@richiejp.com>

* feat(realtime): explicit formally-verified state machines + parakeet streaming driver

The realtime API had several implicit state machines whose state was inferred
from scattered booleans, channels, and five separate mutexes, leaving
illegal/inconsistent states reachable. Make them explicit and keep the
implementation in step with a formal design; rework the parakeet streaming
backend along the same lines.

Realtime state machines (M1-M5). Each is a sealed sum-type State/Event/Effect
with a total, pure Next(state,event)->(state,[]effect) behind a single-writer
Coordinator:

  M1 conncoord    connection lifecycle: VAD toggle + once-only teardown
                  (replaces vadServerStarted + a `done` channel closed from
                  two sites).
  M2 turncoord    turn detection: collapses speechStarted and the live-stream
                  "turn open" flag into one state, so discardTurn can no longer
                  desync them and suppress the next onset.
  M3 respcoord    response coordination: serializes the dual-writer
                  start/cancel so at most one response is live; one
                  response.done per response.create.
  M4 compactcoord conversation compaction: single-flight (replaces the
                  `compacting atomic.Bool` CAS).
  M5 ttscoord     TTS pipeline: open->closing->closed, idempotent wait(),
                  rejects enqueue-after-close (was a silent drop).

The Coordinator/Sink/Next plumbing — only the sealed types and Next differed
per machine — is extracted once into core/http/endpoints/openai/coordinator as
a generic Coordinator[S,E,F]; each machine keeps its public API via type
aliases, so no sink, call-site, or test moved.

Hierarchy. session_lifecycle.fizz models M1 as the parent region with its
children (M2/M3/M4) as one statechart and asserts ChildrenDieWithParent (conn
torn => all children terminal, none start after teardown). respcoord and
compactcoord gain an absorbing Terminated state + Shutdown event; conncoord's
teardown drives the children terminal. This closes a compaction teardown gap: a
fire-and-forget compaction could outlive a torn session — compactionSink now
takes a session-scoped cancellable context + WaitGroup and joins the in-flight
summarize+evict on shutdown.

Formal verification. formal-verification/ holds one authoritative FizzBee spec
per machine plus the composition spec, each with an always-assertion and a
documented one-line edit that makes the checker fail (verified non-vacuous).
scripts/realtime-conformance.sh is fail-closed: all Go conformance suites under
-race AND a model-check of every .fizz spec; a missing FizzBee is a hard error
(only the loud REALTIME_CONFORMANCE_SKIP_FIZZBEE=1 bypasses it, never in CI).
FizzBee is pinned by sha256 and installed via scripts/install-fizzbee.sh into
.tools/ (gitignored). Wired as make test-realtime-conformance, a CI workflow,
and a pre-commit path filter. Go conformance tests are Ginkgo/Gomega (per the
repo's forbidigo lint): transition tables + fixed-seed property walks +
concurrent/-race specs, no rapid dependency. Design map:
docs/design/realtime-state-machines.md.

Parakeet streaming backend. The same treatment applied to the parakeet-cpp
streaming paths:
- AudioTranscriptionStream returns codes.Unimplemented for non-streaming models
  instead of decoding offline and emitting it as one delta + final. A client
  that asked for streaming learns the model cannot stream rather than receiving
  a batch result shaped like a stream. New grpcerrors.StreamTranscriptionUnsupported
  carries that signal; the HTTP /v1/audio/transcriptions stream path surfaces it
  as an SSE error event. Mirrors AudioTranscriptionLive, which already did this.
- utteranceBoundary (boundary.go): a single definition of the end-of-utterance
  latch, replacing three open-coded finalEou toggles. Modelled as a two-valued
  type so illegal states are unrepresentable.
- Shared decode driver (driver.go): streamFeedResult (one per-feed event) +
  feedChunk (hides the ABI v4 JSON vs text-only split) + feedSlices + flushTail.
  The feed loop is written once.
- AudioTranscriptionLive becomes a bidi adapter: it streams the per-feed
  {delta,eou,eob,words} the realtime turn detector consumes and a terminal
  FinalResult carrying only Text. Segments/duration/eou are offline-only and no
  longer produced (nor read) on the live path; liveTraceState drops the terminal
  eou and keeps the per-feed eou_events count.
- AudioTranscriptionStream + streamJSON merge into one driver-based function;
  streamSegmenter is generalized to the unified event with a text-only fallback
  that preserves the legacy (no-words) library's per-utterance segmentation.

Verified: build/vet/gofumpt clean, golangci-lint 0 issues, all coordinator and
parakeet packages under -race, the fail-closed conformance gate green, and
make test-realtime (12 e2e WS+WebRTC).

Assisted-by: Claude:claude-opus-4-8 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>

---------

Signed-off-by: Richard Palethorpe <io@richiejp.com>
2026-06-30 09:01:22 +02:00

84 lines
4.0 KiB
Bash
Executable File

#!/bin/sh
# realtime-conformance: verify the realtime state-machine implementations conform
# to their formal designs. See docs/design/realtime-state-machines.md (Part 6).
#
# Two layers, BOTH required by default -- this gate is FAIL-CLOSED:
# 1. Go-native conformance: the respcoord + turncoord transition tables +
# Ginkgo/Gomega seeded property tests under the race detector (checks the
# implementation).
# 2. FizzBee model check of the authoritative .fizz specs (checks the design,
# and is the thing that makes the design authoritative).
#
# A missing FizzBee is a HARD FAILURE, not a skip -- otherwise verification
# silently evaporates the moment the tool is inconvenient, which defeats the
# whole point. The pinned binary installs reproducibly via scripts/install-fizzbee.sh
# (checksums in formal-verification/fizzbee.sha256), so "couldn't install" is not a
# reason to skip; fix the install. The ONLY way to skip is the explicit, loud
# REALTIME_CONFORMANCE_SKIP_FIZZBEE=1 opt-out, for the rare case of hacking on
# unrelated code locally -- never in CI.
#
# POSIX sh (no bashisms): the project requires tooling scripts to be portable.
set -eu
ROOT=$(CDPATH= cd "$(dirname "$0")/.." && pwd)
GOCMD=${GOCMD:-go}
SPEC_DIR="$ROOT/formal-verification"
echo "==> [1/2] Go conformance (coordinator, respcoord, turncoord, conncoord, compactcoord, ttscoord) with -race"
"$GOCMD" test -race -count=1 \
"$ROOT/core/http/endpoints/openai/coordinator/..." \
"$ROOT/core/http/endpoints/openai/respcoord/..." \
"$ROOT/core/http/endpoints/openai/turncoord/..." \
"$ROOT/core/http/endpoints/openai/conncoord/..." \
"$ROOT/core/http/endpoints/openai/compactcoord/..." \
"$ROOT/core/http/endpoints/openai/ttscoord/..."
# Locate the FizzBee CLI wrapper. install-fizzbee.sh publishes a stable symlink
# at .tools/fizzbee/fizz; otherwise fall back to `fizz` on PATH. FIZZBEE_BIN
# overrides both. NOTE: this must be the `fizz` WRAPPER (which runs the bundled
# parser then the checker), not the raw `fizzbee` binary, and its sibling
# parser/ + fizz.env must be intact.
FIZZBEE_BIN=${FIZZBEE_BIN:-}
if [ -z "$FIZZBEE_BIN" ]; then
if [ -x "$ROOT/.tools/fizzbee/fizz" ]; then
FIZZBEE_BIN="$ROOT/.tools/fizzbee/fizz"
elif command -v fizz >/dev/null 2>&1; then
FIZZBEE_BIN=fizz
fi
fi
echo "==> [2/2] FizzBee model check of authoritative specs"
if [ -n "$FIZZBEE_BIN" ] && { [ -x "$FIZZBEE_BIN" ] || command -v "$FIZZBEE_BIN" >/dev/null 2>&1; }; then
for spec in "$SPEC_DIR"/*.fizz; do
echo " checking $spec"
# CLI is `fizz [flags] <spec.fizz>` (default = exhaustive BFS); there is
# NO `run` subcommand. The checker may print FAILED/DEADLOCK but still
# exit 0, so detect violations from the output as well as the exit code.
set +e
out=$("$FIZZBEE_BIN" "$spec" 2>&1)
rc=$?
set -e
printf '%s\n' "$out"
if [ "$rc" -ne 0 ]; then
echo "ERROR: FizzBee exited $rc on $spec" >&2
exit 1
fi
if printf '%s\n' "$out" | grep -qE '^(FAILED|DEADLOCK)'; then
echo "ERROR: FizzBee reported an invariant/deadlock violation in $spec" >&2
exit 1
fi
done
echo "==> realtime-conformance OK (Go + FizzBee)"
elif [ "${REALTIME_CONFORMANCE_SKIP_FIZZBEE:-0}" = "1" ]; then
echo " !! FizzBee model check EXPLICITLY SKIPPED (REALTIME_CONFORMANCE_SKIP_FIZZBEE=1)" >&2
echo " !! The authoritative design was NOT verified this run. Do not use in CI." >&2
echo "==> realtime-conformance INCOMPLETE (Go only; design check skipped by request)"
else
echo "ERROR: FizzBee not found -- the authoritative design cannot be verified." >&2
echo " This gate is fail-closed; verification is not optional." >&2
echo " Install the pinned, checksum-verified binary: make install-fizzbee" >&2
echo " (see formal-verification/README.md). To deliberately skip locally while" >&2
echo " hacking on unrelated code, set REALTIME_CONFORMANCE_SKIP_FIZZBEE=1." >&2
exit 1
fi