Add SayWithVoice to yaegi symbols

README.md

@@ -8,13 +8,15 @@ Personal voice assistant written in go using DeepSpeech. This was mainly created
 
 The prerequisites for trident are:
 - libdeepspeech and DeepSpeech models (speech to text)
-- mimic (text to speech)
+- flite (text to speech)
 - go
 
 `libdeepspeech` along with its models can be found in [its github releases](https://github.com/mozilla/DeepSpeech/releases/). Be sure to download the `native_client` tarball for your platform and the `.pbmm` and `.scorer` files.
 
-`mimic` can be installed via an [install script](https://github.com/MycroftAI/mycroft-core/blob/dev/scripts/install-mimic.sh) or its AUR package if using Arch Linux or its derivatives.
-
+`flite` can be installed via your distribution's repositories:
+- Debian/Ubuntu: `sudo apt install flite-dev`
+- Fedora: `sudo dnf install flite-devel`
+- Arch: `sudo pacman -S flite festival-us`
 ### Installation
 
 Move the previously downloaded models (`.pbmm` and `.scorer`) into this repo as `deepspeech.pbmm` and `deepspeech.scorer`. Then, follow the next steps.

audio.go

@@ -84,11 +84,16 @@ func playActivationTone(ctx *malgo.AllocatedContext) error {
 
 	// Create new channel waiting for completion
 	done := make(chan bool)
+	doneVar := false
 	onSamples := func(output, _ []byte, _ uint32) {
 		// Read as much audio into output as will fit
 		n, err := io.ReadFull(wavReader, output)
 		// If error occurred or no bytes read
-		if err != nil || n == 0 {
+		if !doneVar && (err != nil || n == 0) {
+			if *verbose {
+				log.Debug().Msg("Sample output complete")
+			}
+			doneVar = true
 			// Signal completion
 			done <- true
 		}

config.go

@@ -93,4 +93,4 @@ func configEnv() (gopath, configDir, execDir, confPath string) {
 	}
 	// Return all variables
 	return
-}
+}

go.mod

@@ -4,6 +4,7 @@ go 1.16
 
 require (
 	github.com/asticode/go-astideepspeech v0.10.0
+	github.com/gen2brain/flite-go v0.0.0-20170519100317-f4df2119132c
 	github.com/gen2brain/malgo v0.10.29
 	github.com/pelletier/go-toml v1.9.0
 	github.com/rs/zerolog v1.21.0

go.sum

@@ -5,6 +5,8 @@ github.com/coreos/go-systemd v0.0.0-20190321100706-95778dfbb74e/go.mod h1:F5haX7
 github.com/cryptix/wav v0.0.0-20180415113528-8bdace674401/go.mod h1:knK8fd+KPlGGqSUWogv1DQzGTwnfUvAi0cIoWyOG7+U=
 github.com/davecgh/go-spew v1.1.0 h1:ZDRjVQ15GmhC3fiQ8ni8+OwkZQO4DARzQgrnXU1Liz8=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/gen2brain/flite-go v0.0.0-20170519100317-f4df2119132c h1:JBlwZJSYopoPXh0dLN9GGw750uhU08VjKKpl+uX5pE4=
+github.com/gen2brain/flite-go v0.0.0-20170519100317-f4df2119132c/go.mod h1:Wv0H30ZpZPf4CrBNqgiG2S4G0CDtZWS2i87JnPtv9LI=
 github.com/gen2brain/malgo v0.10.29 h1:bTYiUTUKJsEomNby+W0hgyLrOttUXIk4lTEnKA54iqM=
 github.com/gen2brain/malgo v0.10.29/go.mod h1:zHSUNZAXfCeNsZou0RtQ6Zk7gDYLIcKOrUWtAdksnEs=
 github.com/pelletier/go-toml v1.9.0 h1:NOd0BRdOKpPf0SxkL3HxSQOG7rNh+4kl6PHcBPFs7Q0=

main.go

@@ -45,6 +45,7 @@ func main() {
 	// Define and parse command line flags
 	tfLogLevel := flag.Int("tf-log-level", 2, "Log level for TensorFlow")
 	verbose = flag.BoolP("verbose", "v", false, "Log more events")
+	showDecode := flag.BoolP("show-decode", "d", false, "Show text to speech decodes")
 	configPath := flag.StringP("config", "c", confPath, "Location of trident TOML config")
 	modelPath := flag.StringP("model", "m", filepath.Join(execDir, "deepspeech.pbmm"), "Path to DeepSpeech model")
 	scorerPath := flag.StringP("scorer", "s", filepath.Join(execDir, "deepspeech.scorer"), "Path to DeepSpeech scorer")
@@ -207,7 +208,7 @@ func main() {
 	// Create goroutine to clean stream every minute
 	go func() {
 		for {
-			time.Sleep(time.Minute)
+			time.Sleep(20 * time.Second)
 			// Lock mutex of stream
 			safeStream.Lock()
 			// Reset stream and buffer
@@ -223,7 +224,7 @@ func main() {
 	var tts string
 	listenForActivation := true
 	for {
-		time.Sleep(time.Second)
+		time.Sleep(200 * time.Millisecond)
 		// Convert captured raw audio to slice of int16
 		slice, err := convToInt16Slice(captured)
 		if err != nil {
@@ -240,6 +241,9 @@ func main() {
 		if err != nil {
 			log.Fatal().Err(err).Msg("Error intermediate decoding stream")
 		}
+		if *showDecode {
+			log.Debug().Msg("TTS Decode: " + tts)
+		}
 		// If decoded string contains activation phrase and listenForActivation is true
 		if strings.Contains(tts, config.ActivationPhrase) && listenForActivation {
 			// Play activation tone

plugins.go

@@ -100,4 +100,3 @@ func initPlugins(gopath string) map[string]pluginFunc {
 	}
 	return out
 }
-

plugins/shell/shell.go

@@ -21,18 +21,21 @@ package shell
 import (
 	"os"
 	"os/exec"
+	"trident"
 )
 
 func RunPlugin(program string, data map[string]interface{}) {
-	var shell string
-	var ok bool
 	// Attempt to get shell from config, asserting as string
-	shell, ok = data["shell"].(string)
+	shell, ok := data["shell"].(string)
 	// If unsuccessful
 	if !ok {
 		// Set shell to default (/bin/sh)
 		shell = "/bin/sh"
 	}
+	sayOutput, ok := data["sayOutput"].(bool)
+	if !ok {
+		sayOutput = false
+	}
 	// Create command using configured shell or default (/bin/sh)
 	cmd := exec.Command(shell, "-c", program)
 	// Set command environment to system environment
@@ -40,5 +43,8 @@ func RunPlugin(program string, data map[string]interface{}) {
 	// Set command's standard error to system standard error
 	cmd.Stderr = os.Stderr
 	// Run command, ignoring error
-	_ = cmd.Run()
+	output, _ := cmd.Output()
+	if sayOutput {
+		trident.Say(string(output))
+	}
 }

symbols.go

@@ -18,28 +18,28 @@
 package main
 
 import (
+	"github.com/gen2brain/flite-go"
 	"github.com/traefik/yaegi/interp"
-	"os/exec"
 	"reflect"
 )
 
 // Create custom package for trident
 var tridentSymbols = interp.Exports{"trident": {
-	"Say": reflect.ValueOf(Say),
+	"Say":          reflect.ValueOf(Say),
+	"SayWithVoice": reflect.ValueOf(SayWithVoice),
 }}
 
 // Function to say text using mimic text-to-speech
-func Say(text string, args ...string) {
-	// If mimic exists in PATH
-	if _, err := exec.LookPath("mimic"); err == nil {
-		// Set initial argument slice to contain text
-		argSlice := []string{"-t", text}
-		// Add any additional arguments to slice
-		argSlice = append(argSlice, args...)
-		// Create and run command
-		exec.Command("mimic", argSlice...).Run()
-	} else {
-		// If mimic does not exist in PATH, warn user
-		log.Warn().Err(err).Str("text", text).Msg("Cannot perform text to speech")
+func Say(text string) {
+	fliteVoice, _ := flite.VoiceSelect("slt")
+	flite.TextToSpeech(text, fliteVoice, "play")
+}
+
+func SayWithVoice(text, voice string) error {
+	fliteVoice, err := flite.VoiceSelect(voice)
+	if err != nil {
+		return err
 	}
-}
+	flite.TextToSpeech(text, fliteVoice, "play")
+	return nil
+}