Nik Afiq 5238298b55
Some checks failed
CI / test (push) Successful in 6s
CI / build-ai-gateway (push) Failing after 28s
CI / build-ha-gateway (push) Failing after 24s
CI / build-discord-bot (push) Failing after 29s
Add TTS model components and inference server
- Implemented core model components in `modules.py` including various convolutional layers and normalization techniques.
- Added transformation functions in `transforms.py` for piecewise rational quadratic transformations.
- Created utility functions in `utils.py` for checkpoint management, logging, and hyperparameter handling.
- Introduced monotonic alignment functionality with Cython optimization in `monotonic_align`.
- Developed a minimal inference server in `server.py` to handle synthesis requests.
- Updated requirements to include necessary dependencies for Cython and scipy.
2026-07-24 22:38:36 +09:00

160 lines
4.6 KiB
Go

package main
import (
"context"
"crypto/tls"
"crypto/x509"
"fmt"
"net"
"os"
"os/signal"
"path/filepath"
"syscall"
"github.com/joho/godotenv"
"go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc"
"google.golang.org/grpc"
"google.golang.org/grpc/credentials"
"google.golang.org/grpc/health"
grpc_health_v1 "google.golang.org/grpc/health/grpc_health_v1"
"google.golang.org/grpc/reflection"
ttsv1 "gitea.nik4nao.com/nik/home-services/gen/tts/v1"
grpcadapter "gitea.nik4nao.com/nik/home-services/tts-gateway/internal/adapters/primary/grpc"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/adapters/secondary/ffmpeg"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/adapters/secondary/inferencesidecar"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/adapters/secondary/jtalk"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/app"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/config"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/logger"
"gitea.nik4nao.com/nik/home-services/tts-gateway/internal/telemetry"
)
// MEMO: auth is not implemented - see ha-gateway/cmd/gateway/main.go's memo
// for the same options (shared API key vs mTLS) before exposing this
// service to any untrusted network.
// version is set at build time via -ldflags "-X main.version=<tag>".
var version = "dev"
func main() {
_ = godotenv.Load()
cfg, err := config.Load()
if err != nil {
os.Stderr.WriteString("config error: " + err.Error() + "\n")
os.Exit(1)
}
log := logger.New(cfg.LogFormat, cfg.LogLevel)
log.Info("starting tts-gateway",
"version", version,
"grpc_port", cfg.GRPCPort,
"tls_dir", cfg.TLSDir,
"otel_endpoint", cfg.OTELEndpoint,
"log_level", cfg.LogLevel,
"log_format", cfg.LogFormat,
"inference_sidecar_addr", cfg.InferenceSidecarAddr,
)
ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGTERM, syscall.SIGINT)
defer stop()
ctx = logger.WithLogger(ctx, log)
shutdown, err := telemetry.Setup(ctx, "tts-gateway", version, cfg)
if err != nil {
log.Error("telemetry setup failed", "err", err)
os.Exit(1)
}
if cfg.OTELEndpoint != "" {
log.Info("telemetry enabled", "endpoint", cfg.OTELEndpoint)
} else {
log.Debug("telemetry disabled")
}
normalizer, err := jtalk.NewClient(cfg)
if err != nil {
log.Error("open_jtalk client setup failed", "err", err)
os.Exit(1)
}
engine := inferencesidecar.NewClient(cfg.InferenceSidecarAddr)
encoder := ffmpeg.NewEncoder()
ttsApp := app.NewTTSApp(normalizer, engine, encoder)
serverOpts := []grpc.ServerOption{
grpc.StatsHandler(otelgrpc.NewServerHandler()),
grpc.ChainUnaryInterceptor(grpcadapter.LoggingUnaryInterceptor(log)),
}
if cfg.TLSDir != "" {
creds, err := loadServerCredentials(cfg.TLSDir)
if err != nil {
log.Error("load mTLS credentials failed", "tls_dir", cfg.TLSDir, "err", err)
os.Exit(1)
}
serverOpts = append(serverOpts, grpc.Creds(creds))
log.Info("mTLS enabled", "tls_dir", cfg.TLSDir)
} else {
log.Info("mTLS disabled")
}
srv := grpc.NewServer(serverOpts...)
healthSrv := health.NewServer()
healthSrv.SetServingStatus("", grpc_health_v1.HealthCheckResponse_SERVING)
ttsv1.RegisterTTSServiceServer(srv, grpcadapter.NewTTSGRPC(ttsApp))
grpc_health_v1.RegisterHealthServer(srv, healthSrv)
reflection.Register(srv)
lis, err := net.Listen("tcp", ":"+cfg.GRPCPort)
if err != nil {
log.Error("listen failed", "err", err)
os.Exit(1)
}
go func() {
log.Info("tts-gateway listening", "addr", lis.Addr().String())
if err := srv.Serve(lis); err != nil {
log.Error("serve failed", "err", err)
}
}()
<-ctx.Done()
log.Info("shutdown signal received, draining")
healthSrv.SetServingStatus("", grpc_health_v1.HealthCheckResponse_NOT_SERVING)
srv.GracefulStop()
log.Info("shutdown complete")
if err := shutdown(context.Background()); err != nil {
log.Error("telemetry shutdown error", "err", err)
}
}
func loadServerCredentials(tlsDir string) (credentials.TransportCredentials, error) {
cert, err := tls.LoadX509KeyPair(
filepath.Join(tlsDir, "tls.crt"),
filepath.Join(tlsDir, "tls.key"),
)
if err != nil {
return nil, fmt.Errorf("load server key pair: %w", err)
}
caPEM, err := os.ReadFile(filepath.Join(tlsDir, "ca.crt"))
if err != nil {
return nil, fmt.Errorf("read client CA: %w", err)
}
clientCAs := x509.NewCertPool()
if !clientCAs.AppendCertsFromPEM(caPEM) {
return nil, fmt.Errorf("append client CA: invalid PEM")
}
return credentials.NewTLS(&tls.Config{
Certificates: []tls.Certificate{cert},
ClientCAs: clientCAs,
ClientAuth: tls.RequireAndVerifyClientCert,
MinVersion: tls.VersionTLS13,
}), nil
}