Files
LocalAI/backend/go/parakeet-cpp/segments_test.go
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

155 lines
6.1 KiB
Go

package main
import (
pb "github.com/mudler/LocalAI/pkg/grpc/proto"
. "github.com/onsi/ginkgo/v2"
. "github.com/onsi/gomega"
)
func tw(text string, start, end float64) transcriptWord {
return transcriptWord{W: text, Start: start, End: end}
}
var _ = Describe("splitWordsIntoSegments (NeMo get_segment_offsets parity)", func() {
seps := []rune{'.', '?', '!'}
It("splits on sentence-ending punctuation, including the delimiter word", func() {
words := []transcriptWord{tw("hello", 0, 0.4), tw("world.", 0.4, 0.8), tw("bye", 1.0, 1.3)}
segs := splitWordsIntoSegments(words, seps, 0)
Expect(segs).To(HaveLen(2))
Expect(segs[0]).To(HaveLen(2))
Expect(segs[0][1].W).To(Equal("world."))
Expect(segs[1]).To(HaveLen(1))
Expect(segs[1][0].W).To(Equal("bye"))
})
It("keeps a single segment with no terminal punctuation and gap off", func() {
words := []transcriptWord{tw("a", 0, 0.2), tw("b", 0.2, 0.4), tw("c", 5.0, 5.2)}
segs := splitWordsIntoSegments(words, seps, 0)
Expect(segs).To(HaveLen(1))
})
It("splits on the gap rule when enabled, the gapped word starting the next segment", func() {
words := []transcriptWord{tw("a", 0, 0.2), tw("b", 0.2, 0.4), tw("c", 5.0, 5.2)}
segs := splitWordsIntoSegments(words, seps, 1.0) // c is 4.6s after b
Expect(segs).To(HaveLen(2))
Expect(segs[0]).To(HaveLen(2)) // a b
Expect(segs[1]).To(HaveLen(1)) // c
Expect(segs[1][0].W).To(Equal("c"))
})
It("checks the gap rule before punctuation (NeMo elif order)", func() {
// "b." would terminate, but c is far after it -> gap closes [a b.] at b.
words := []transcriptWord{tw("a", 0, 0.2), tw("b.", 0.2, 0.4), tw("c", 9.0, 9.2)}
segs := splitWordsIntoSegments(words, seps, 1.0)
Expect(segs).To(HaveLen(2))
Expect(segs[0]).To(HaveLen(2))
Expect(segs[1][0].W).To(Equal("c"))
})
It("still splits on punctuation when the gap rule is enabled but does not fire", func() {
words := []transcriptWord{tw("hi.", 0, 0.4), tw("bye", 0.4, 0.8)}
segs := splitWordsIntoSegments(words, seps, 5.0) // gap never reached
Expect(segs).To(HaveLen(2))
Expect(segs[0][0].W).To(Equal("hi."))
})
It("returns nothing for empty input", func() {
Expect(splitWordsIntoSegments(nil, seps, 0)).To(BeEmpty())
})
})
var _ = Describe("transcriptResultFromDoc (multi-segment)", func() {
doc := transcriptJSON{
Text: "hello world. bye now",
FrameSec: 0.08,
Words: []transcriptWord{
{W: "hello", Start: 0.0, End: 0.4},
{W: "world.", Start: 0.4, End: 0.8},
{W: "bye", Start: 1.0, End: 1.3},
{W: "now", Start: 1.3, End: 1.6},
},
Tokens: []transcriptToken{{ID: 1, T: 0.1}, {ID: 2, T: 0.5}, {ID: 3, T: 1.1}, {ID: 4, T: 1.4}},
}
It("emits one segment per punctuation-delimited group with start/end", func() {
res := transcriptResultFromDoc(doc, &pb.TranscriptRequest{}, 0)
Expect(res.Segments).To(HaveLen(2))
Expect(res.Segments[0].Text).To(Equal("hello world."))
Expect(res.Segments[0].Start).To(Equal(int64(0)))
Expect(res.Segments[0].End).To(Equal(secondsToNanos(0.8)))
Expect(res.Segments[1].Text).To(Equal("bye now"))
Expect(res.Segments[1].Start).To(Equal(secondsToNanos(1.0)))
Expect(res.Segments[1].Id).To(Equal(int32(1)))
})
It("assigns tokens to the segment whose time window contains them", func() {
res := transcriptResultFromDoc(doc, &pb.TranscriptRequest{}, 0)
Expect(res.Segments[0].Tokens).To(Equal([]int32{1, 2}))
Expect(res.Segments[1].Tokens).To(Equal([]int32{3, 4}))
})
It("attaches per-segment words only when word granularity requested", func() {
plain := transcriptResultFromDoc(doc, &pb.TranscriptRequest{}, 0)
Expect(plain.Segments[0].Words).To(BeEmpty())
withWords := transcriptResultFromDoc(doc, &pb.TranscriptRequest{TimestampGranularities: []string{"word"}}, 0)
Expect(withWords.Segments[0].Words).To(HaveLen(2))
})
It("falls back to a single text segment when there are no words", func() {
res := transcriptResultFromDoc(transcriptJSON{Text: "hi"}, &pb.TranscriptRequest{}, 0)
Expect(res.Segments).To(HaveLen(1))
Expect(res.Segments[0].Text).To(Equal("hi"))
})
})
var _ = Describe("streaming segment assembly", func() {
It("closes a segment with start/end from its words on EOU", func() {
acc := &streamSegmenter{}
acc.add(streamFeedResult{Delta: "hello world", Eou: true, Words: []transcriptWord{
{W: "hello", Start: 0.0, End: 0.4}, {W: "world", Start: 0.4, End: 0.9},
}})
segs := acc.segments()
Expect(segs).To(HaveLen(1))
Expect(segs[0].Text).To(Equal("hello world"))
Expect(segs[0].Start).To(Equal(int64(0)))
Expect(segs[0].End).To(Equal(secondsToNanos(0.9)))
})
It("buffers words across feeds until EOU", func() {
acc := &streamSegmenter{}
acc.add(streamFeedResult{Delta: "hi", Words: []transcriptWord{{W: "hi", Start: 0, End: 0.3}}})
Expect(acc.segments()).To(BeEmpty())
acc.add(streamFeedResult{Delta: "there", Eou: true, Words: []transcriptWord{{W: "there", Start: 0.3, End: 0.7}}})
Expect(acc.segments()).To(HaveLen(1))
Expect(acc.segments()[0].Text).To(Equal("hi there"))
})
// ABI v5 split <EOB> (backchannel) out of the "eou" flag into its own "eob"
// field; a backchannel must still close the segment as it did in v4.
It("closes a segment on EOB (backchannel) too", func() {
acc := &streamSegmenter{}
acc.add(streamFeedResult{Delta: "uh huh", Eob: true, Words: []transcriptWord{
{W: "uh", Start: 0.0, End: 0.2}, {W: "huh", Start: 0.2, End: 0.5},
}})
segs := acc.segments()
Expect(segs).To(HaveLen(1))
Expect(segs[0].Text).To(Equal("uh huh"))
Expect(segs[0].End).To(Equal(secondsToNanos(0.5)))
})
// Older text-only libparakeet.so: no per-word timings, so a segment is cut
// from the delta text on each <EOU>/<EOB> (no timestamps), one per utterance.
It("falls back to text segments when the feed carries no words", func() {
acc := &streamSegmenter{}
acc.add(streamFeedResult{Delta: "first turn", Eou: true})
acc.add(streamFeedResult{Delta: "second turn", Eou: true})
segs := acc.segments()
Expect(segs).To(HaveLen(2))
Expect(segs[0].Text).To(Equal("first turn"))
Expect(segs[1].Text).To(Equal("second turn"))
Expect(segs[0].Start).To(Equal(int64(0)), "no per-word timing on the text path")
Expect(segs[0].End).To(Equal(int64(0)))
})
})