Fix building of stt

cmd/stt/main.go

@@ -13,7 +13,6 @@ import (
 	"github.com/facebookincubator/go-belt/tool/logger"
 	"github.com/facebookincubator/go-belt/tool/logger/implementation/logrus"
 	"github.com/lazybeaver/entropy"
-	syswhisper "github.com/mutablelogic/go-whisper/sys/whisper"
 	"github.com/spf13/pflag"
 	"github.com/xaionaro-go/observability"
 	"github.com/xaionaro-go/speech/pkg/speech"
@@ -34,7 +33,7 @@ func main() {
 	loggerLevel := logger.LevelWarning
 	pflag.Var(&loggerLevel, "log-level", "Log level")
 	langFlag := pflag.String("language", "en-US", "")
-	alignmentAheadPresentFlag := types.AlignmentAheadsPreset(syswhisper.AlignmentAheadsPresetNone)
+	alignmentAheadPresentFlag := types.AlignmentAheadsPreset(speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetNone)
 	pflag.Var(&alignmentAheadPresentFlag, "alignment-aheads-preset", "")
 	gpuFlag := pflag.Int("gpu", -1, "")
 	useGPUFlag := pflag.Bool("use-gpu", true, "")
@@ -94,7 +93,7 @@ func main() {
 			Backend: &speechtotext_grpc.NewContextRequest_Whisper{
 				Whisper: &speechtotext_grpc.WhisperOptions{
 					SamplingStrategy:      goconv.SamplingStrategyToGRPC(types.SamplingStrategyGreedy),
-					AlignmentAheadsPreset: goconv.AlignmentAheadsPresetToGRPC(syswhisper.AlignmentAheadsPreset(alignmentAheadPresentFlag)),
+					AlignmentAheadsPreset: speechtotext_grpc.WhisperAlignmentAheadsPreset(alignmentAheadPresentFlag),
 				},
 			},
 		})
@@ -106,7 +105,7 @@ func main() {
 			speech.Language(*langFlag),
 			types.SamplingStrategyGreedy,
 			*shouldTranslateFlag,
-			whisper.AlignmentAheadsPreset(alignmentAheadPresentFlag),
+			alignmentAheadPresentFlag,
 			*vadThreshold,
 			opts...,
 		)

pkg/speech/speechtotext/implementations/whisper/alignment_aheads_preset.go

@@ -0,0 +1,44 @@
+package whisper
+
+import (
+	"github.com/mutablelogic/go-whisper/sys/whisper"
+	"github.com/xaionaro-go/speech/pkg/speech/speechtotext/implementations/whisper/types"
+	"github.com/xaionaro-go/speech/pkg/speech/speechtotext/server/proto/go/speechtotext_grpc"
+)
+
+type AlignmentAheadsPreset types.AlignmentAheadsPreset
+
+// String just implements fmt.Stringer, flag.Value and pflag.Value.
+func (p AlignmentAheadsPreset) ToWhisper() whisper.AlignmentAheadsPreset {
+	switch speechtotext_grpc.WhisperAlignmentAheadsPreset(p) {
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetNone:
+		return whisper.AlignmentAheadsPresetNone
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetNTopMost:
+		return whisper.AlignmentAheadsPresetNTopMost
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetCustom:
+		return whisper.AlignmentAheadsPresetCustom
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetTinyEn:
+		return whisper.AlignmentAheadsPresetTinyEn
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetTiny:
+		return whisper.AlignmentAheadsPresetTiny
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetBaseEn:
+		return whisper.AlignmentAheadsPresetBaseEn
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetBase:
+		return whisper.AlignmentAheadsPresetBase
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetSmallEn:
+		return whisper.AlignmentAheadsPresetSmallEn
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetSmall:
+		return whisper.AlignmentAheadsPresetSmall
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetMediumEn:
+		return whisper.AlignmentAheadsPresetMediumEn
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetMedium:
+		return whisper.AlignmentAheadsPresetMedium
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetLargeV1:
+		return whisper.AlignmentAheadsPresetLargeV1
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetLargeV2:
+		return whisper.AlignmentAheadsPresetLargeV2
+	case speechtotext_grpc.WhisperAlignmentAheadsPreset_WhisperAlignmentAheadsPresetLargeV3:
+		return whisper.AlignmentAheadsPresetLargeV3
+	}
+	panic(p)
+}

pkg/speech/speechtotext/implementations/whisper/speech_to_text.go

@@ -99,7 +99,7 @@ func New(
 		params.SetFlashAttn(*cfg.FlashAttn)
 	}
 	params.SetTokenTimestamps(false)
-	params.SetDTWAheadsPreset(whisper.AlignmentAheadsPreset(alignmentAheadPreset))
+	params.SetDTWAheadsPreset(AlignmentAheadsPreset(alignmentAheadPreset).ToWhisper())
 	whisper.Whisper_log_set(func(level whisper.LogLevel, text string) {
 		logger.FromCtx(ctx).Log(logLevelFromWhisper(level), text)
 	})