mirror of
https://github.com/mudler/LocalAI.git
synced 2026-06-30 19:37:00 -04:00
* 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>
238 lines
7.0 KiB
Go
238 lines
7.0 KiB
Go
package base
|
|
|
|
// This is a wrapper to satisfy the GRPC service interface
|
|
// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"os"
|
|
|
|
"github.com/mudler/LocalAI/pkg/grpc/grpcerrors"
|
|
pb "github.com/mudler/LocalAI/pkg/grpc/proto"
|
|
gopsutil "github.com/shirou/gopsutil/v3/process"
|
|
)
|
|
|
|
// Base is a base class for all backends to implement
|
|
// Note: the backends that does not support multiple requests
|
|
// should use SingleThread instead
|
|
type Base struct {
|
|
}
|
|
|
|
func (llm *Base) Locking() bool {
|
|
return false
|
|
}
|
|
|
|
func (llm *Base) Lock() {
|
|
panic("not implemented")
|
|
}
|
|
|
|
func (llm *Base) Unlock() {
|
|
panic("not implemented")
|
|
}
|
|
|
|
func (llm *Base) Busy() bool {
|
|
return false
|
|
}
|
|
|
|
func (llm *Base) Load(opts *pb.ModelOptions) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) Predict(opts *pb.PredictOptions) (string, error) {
|
|
return "", fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) PredictStream(opts *pb.PredictOptions, results chan string) error {
|
|
close(results)
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) Embeddings(opts *pb.PredictOptions) ([]float32, error) {
|
|
return []float32{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) GenerateImage(*pb.GenerateImageRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) GenerateVideo(*pb.GenerateVideoRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioTranscription(context.Context, *pb.TranscriptRequest) (pb.TranscriptResult, error) {
|
|
return pb.TranscriptResult{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioTranscriptionStream(context.Context, *pb.TranscriptRequest, chan *pb.TranscriptStreamResponse) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) TTS(*pb.TTSRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) TTSStream(*pb.TTSRequest, chan []byte) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) SoundGeneration(*pb.SoundGenerationRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) Detect(*pb.DetectOptions) (pb.DetectResponse, error) {
|
|
return pb.DetectResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) Depth(*pb.DepthRequest) (pb.DepthResponse, error) {
|
|
return pb.DepthResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) FaceVerify(*pb.FaceVerifyRequest) (pb.FaceVerifyResponse, error) {
|
|
return pb.FaceVerifyResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) FaceAnalyze(*pb.FaceAnalyzeRequest) (pb.FaceAnalyzeResponse, error) {
|
|
return pb.FaceAnalyzeResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) VoiceVerify(*pb.VoiceVerifyRequest) (pb.VoiceVerifyResponse, error) {
|
|
return pb.VoiceVerifyResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) VoiceAnalyze(*pb.VoiceAnalyzeRequest) (pb.VoiceAnalyzeResponse, error) {
|
|
return pb.VoiceAnalyzeResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) VoiceEmbed(*pb.VoiceEmbedRequest) (pb.VoiceEmbedResponse, error) {
|
|
return pb.VoiceEmbedResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) Diarize(*pb.DiarizeRequest) (pb.DiarizeResponse, error) {
|
|
return pb.DiarizeResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) SoundDetection(context.Context, *pb.SoundDetectionRequest) (*pb.SoundDetectionResponse, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) TokenizeString(opts *pb.PredictOptions) (pb.TokenizationResponse, error) {
|
|
return pb.TokenizationResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) ModelMetadata(opts *pb.ModelOptions) (*pb.ModelMetadataResponse, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
// backends may wish to call this to capture the gopsutil info, then enhance with additional memory usage details?
|
|
func (llm *Base) Status() (pb.StatusResponse, error) {
|
|
return pb.StatusResponse{
|
|
Memory: memoryUsage(),
|
|
}, nil
|
|
}
|
|
|
|
func (llm *Base) StoresSet(*pb.StoresSetOptions) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StoresGet(*pb.StoresGetOptions) (pb.StoresGetResult, error) {
|
|
return pb.StoresGetResult{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StoresDelete(*pb.StoresDeleteOptions) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StoresFind(*pb.StoresFindOptions) (pb.StoresFindResult, error) {
|
|
return pb.StoresFindResult{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) VAD(*pb.VADRequest) (pb.VADResponse, error) {
|
|
return pb.VADResponse{}, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioEncode(*pb.AudioEncodeRequest) (*pb.AudioEncodeResult, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioDecode(*pb.AudioDecodeRequest) (*pb.AudioDecodeResult, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioTransform(*pb.AudioTransformRequest) (*pb.AudioTransformResult, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioTransformStream(in <-chan *pb.AudioTransformFrameRequest, out chan<- *pb.AudioTransformFrameResponse) error {
|
|
close(out)
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) AudioTranscriptionLive(in <-chan *pb.TranscriptLiveRequest, out chan<- *pb.TranscriptLiveResponse) error {
|
|
close(out)
|
|
return grpcerrors.LiveTranscriptionUnsupported("base", "not implemented by this backend")
|
|
}
|
|
|
|
func (llm *Base) AudioToAudioStream(in <-chan *pb.AudioToAudioRequest, out chan<- *pb.AudioToAudioResponse) error {
|
|
close(out)
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) Forward(ctx context.Context, in <-chan *pb.ForwardRequest, out chan<- *pb.ForwardReply) error {
|
|
close(out)
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StartFineTune(*pb.FineTuneRequest) (*pb.FineTuneJobResult, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) FineTuneProgress(*pb.FineTuneProgressRequest, chan *pb.FineTuneProgressUpdate) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StopFineTune(*pb.FineTuneStopRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) ListCheckpoints(*pb.ListCheckpointsRequest) (*pb.ListCheckpointsResponse, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) ExportModel(*pb.ExportModelRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StartQuantization(*pb.QuantizationRequest) (*pb.QuantizationJobResult, error) {
|
|
return nil, fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) QuantizationProgress(*pb.QuantizationProgressRequest, chan *pb.QuantizationProgressUpdate) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func (llm *Base) StopQuantization(*pb.QuantizationStopRequest) error {
|
|
return fmt.Errorf("unimplemented")
|
|
}
|
|
|
|
func memoryUsage() *pb.MemoryUsageData {
|
|
mud := pb.MemoryUsageData{
|
|
Breakdown: make(map[string]uint64),
|
|
}
|
|
|
|
pid := int32(os.Getpid())
|
|
|
|
backendProcess, err := gopsutil.NewProcess(pid)
|
|
|
|
if err == nil {
|
|
memInfo, err := backendProcess.MemoryInfo()
|
|
if err == nil {
|
|
mud.Total = memInfo.VMS // TEST, but rss seems reasonable first guess. Does include swap, but we might care about that.
|
|
mud.Breakdown["gopsutil-RSS"] = memInfo.RSS
|
|
}
|
|
}
|
|
return &mud
|
|
}
|
|
func (llm *Base) Free() error {
|
|
return nil
|
|
}
|