Merge pull request #351 from kercre123/highpass-filter

Remove soxr

chipper/go.mod

@@ -36,7 +36,6 @@ require (
 	github.com/akavel/rsrc v0.10.2 // indirect
 	github.com/alecthomas/template v0.0.0-20160405071501-a0175ee3bccc // indirect
 	github.com/alecthomas/units v0.0.0-20151022065526-2efee857e7cf // indirect
-	github.com/alphacep/vosk-api/go v0.3.50 // indirect
 	github.com/cenkalti/backoff v2.2.1+incompatible // indirect
 	github.com/currantlabs/ble v0.0.0-20171229162446-c1d21c164cf8 // indirect
 	github.com/dchest/jsmin v0.0.0-20220218165748-59f39799265f // indirect
@@ -49,7 +48,6 @@ require (
 	github.com/hashicorp/hcl v1.0.0 // indirect
 	github.com/jamesruan/sodium v0.0.0-20181216154042-9620b83ffeae // indirect
 	github.com/josephspurrier/goversioninfo v1.4.0 // indirect
-	github.com/kercre123/vosk-api v1.0.1 // indirect
 	github.com/kr/text v0.2.0 // indirect
 	github.com/magiconair/properties v1.8.1 // indirect
 	github.com/mattn/go-colorable v0.1.8 // indirect
@@ -67,7 +65,6 @@ require (
 	github.com/spf13/pflag v1.0.5 // indirect
 	github.com/spf13/viper v1.7.1 // indirect
 	github.com/subosito/gotenv v1.2.0 // indirect
-	github.com/zaf/resample v1.5.0 // indirect
 	golang.org/x/image v0.10.0 // indirect
 	golang.org/x/net v0.16.0 // indirect
 	golang.org/x/sys v0.15.0 // indirect

chipper/go.sum

@@ -53,8 +53,6 @@ github.com/alecthomas/template v0.0.0-20160405071501-a0175ee3bccc h1:cAKDfWh5Vpd
 github.com/alecthomas/template v0.0.0-20160405071501-a0175ee3bccc/go.mod h1:LOuyumcjzFXgccqObfd/Ljyb9UuFJ6TxHnclSeseNhc=
 github.com/alecthomas/units v0.0.0-20151022065526-2efee857e7cf h1:qet1QNfXsQxTZqLG4oE62mJzwPIB8+Tee4RNCL9ulrY=
 github.com/alecthomas/units v0.0.0-20151022065526-2efee857e7cf/go.mod h1:ybxpYRFXyAe+OPACYpWeL0wqObRcbAqCMya13uyzqw0=
-github.com/alphacep/vosk-api/go v0.3.50 h1:2vSN41RCU1WdHEqBrhKtTggfKL6Yu5Dmj+urVszwiuw=
-github.com/alphacep/vosk-api/go v0.3.50/go.mod h1:9X8IJsHnFk/b1xyvjlZifo+ZL5VTAx3LW+JQce/eRcA=
 github.com/antihax/optional v1.0.0/go.mod h1:uupD/76wgC+ih3iEmQUL+0Ugr19nfwCT1kdvxnR2qWY=
 github.com/arbovm/levenshtein v0.0.0-20160628152529-48b4e1c0c4d0/go.mod h1:t2tdKJDJF9BV14lnkjHmOQgcvEKgtqs5a1N3LNdJhGE=
 github.com/armon/circbuf v0.0.0-20150827004946-bbbad097214e/go.mod h1:3U/XgcO3hCbHZ8TKRvWD2dDTCfh9M9ya+I9JpbB7O8o=
@@ -138,8 +136,6 @@ github.com/fsnotify/fsnotify v1.4.7/go.mod h1:jwhsz4b93w/PPRr/qN1Yymfu8t87LnFCMo
 github.com/fsnotify/fsnotify v1.4.9 h1:hsms1Qyu0jgnwNXIxa+/V/PDsU6CfLf6CNO8H7IWoS4=
 github.com/fsnotify/fsnotify v1.4.9/go.mod h1:znqG4EE+3YCdAaPaxE2ZRY/06pZUdp0tY4IgpuI1SZQ=
 github.com/fsouza/go-dockerclient v1.6.6/go.mod h1:3/oRIWoe7uT6bwtAayj/EmJmepBjeL4pYvt7ZxC7Rnk=
-github.com/ggerganov/whisper.cpp/bindings/go v0.0.0-20231214200047-940de9dbe9c9 h1:B3bFAoLQ8y4RFNR8A4GjkCiKzuUIwblYUiIgRgWr604=
-github.com/ggerganov/whisper.cpp/bindings/go v0.0.0-20231214200047-940de9dbe9c9/go.mod h1:QIjZ9OktHFG7p+/m3sMvrAJKKdWrr1fZIK0rM6HZlyo=
 github.com/ggerganov/whisper.cpp/bindings/go v0.0.0-20240618151033-bf4cb4abad4e h1:np99/bjGH4/khEujoGbwc0ohMLh32GjovhEv2mJRNfs=
 github.com/ggerganov/whisper.cpp/bindings/go v0.0.0-20240618151033-bf4cb4abad4e/go.mod h1:QIjZ9OktHFG7p+/m3sMvrAJKKdWrr1fZIK0rM6HZlyo=
 github.com/ghodss/yaml v1.0.0/go.mod h1:4dBDuWmgqj2HViK6kFavaiC9ZROes6MMH2rRYeMEF04=
@@ -308,8 +304,6 @@ github.com/julienschmidt/httprouter v1.2.0/go.mod h1:SYymIcj16QtmaHHD7aYtjjsJG7V
 github.com/k0kubun/go-ansi v0.0.0-20180517002512-3bf9e2903213/go.mod h1:vNUNkEQ1e29fT/6vq2aBdFsgNPmy8qMdSay1npru+Sw=
 github.com/karrick/godirwalk v1.8.0/go.mod h1:H5KPZjojv4lE+QYImBI8xVtrBRgYrIVsaRPx4tDPEn4=
 github.com/karrick/godirwalk v1.10.3/go.mod h1:RoGL9dQei4vP9ilrpETWE8CLOZ1kiN0LhBygSwrAsHA=
-github.com/kercre123/vosk-api v1.0.1 h1:D5CeAMNHPj87M9fKrqP+a2gEQefq7sJCpaiuRscbiJY=
-github.com/kercre123/vosk-api v1.0.1/go.mod h1:mJlLhtYS207jVY9QffYGxhX6Up0UfSQ3p0uNbXsf3Zc=
 github.com/kercre123/vosk-api/go v1.0.2 h1:NDJUNv2ddw128amiVZ2xE2gKfKHeBRRhboSh+yiH6Wg=
 github.com/kercre123/vosk-api/go v1.0.2/go.mod h1:oVZG/VFmg23uNDzjShcw7UhZHWYG2zXgBm5FqioE2Ao=
 github.com/kercre123/zeroconf v1.0.1 h1:Mbd8mN6xnNWYIqBN38x3jJjiPP2RmK4orzbGZsa1EOY=
@@ -495,8 +489,6 @@ github.com/yuin/goldmark v1.1.25/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9de
 github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.1.32/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.4.13/go.mod h1:6yULJ656Px+3vBD8DxQVa3kxgyrAnzto9xy5taEt/CY=
-github.com/zaf/resample v1.5.0 h1:c3yumHrV1cJoED8ZY2Ai3cehS8s0mJSroA9/vMaUcho=
-github.com/zaf/resample v1.5.0/go.mod h1:e4yWalfgRccQrnZSrkIxTqmMCOPhTi1xvYpNpRIB13k=
 go.etcd.io/bbolt v1.3.2/go.mod h1:IbVyRI1SCnLcuJnV2u8VeU0CEYM7e686BmAb1XKL+uU=
 go.mongodb.org/mongo-driver v1.4.2/go.mod h1:WcMNYLx/IlOxLe6JRJiv2uXuCz6zBLndR4SoGjYphSc=
 go.opencensus.io v0.21.0/go.mod h1:mSImk1erAIZhrmZN+AvHh14ztQfjbGwt4TtuofqLduU=

chipper/pkg/wirepod/ttr/convert.go

@@ -1,12 +1,8 @@
 package wirepod_ttr
 
-// TODO
 import (
-	"bytes"
 	"encoding/binary"
 	"math"
-
-	"github.com/zaf/resample"
 )
 
 func bytesToInt16s(data []byte) []int16 {
@@ -25,16 +21,20 @@ func int16sToBytes(data []int16) []byte {
 	return bytes
 }
 
-// entry-point (go, not default)
 func downsample24kTo16k(input []byte) [][]byte {
-	iVolBytes := increaseVolume(input, 6)
-	outBytes := downsample24kTo16kLinear(iVolBytes)
+	outBytes := downsample24kTo16kLinear(input)
 	var audioChunks [][]byte
-	// the "s" sounds are harsh. put through filter
-	//filteredBytes := lowPassFilter(outBytes, 4000, 16000)
-	for len(outBytes) >= 1024 {
-		audioChunks = append(audioChunks, outBytes[:1024])
-		outBytes = outBytes[1024:]
+	filteredBytes := lowPassFilter(outBytes, 4000, 16000)
+	iVolBytes := increaseVolume(filteredBytes, 5)
+	for len(iVolBytes) > 0 {
+		if len(iVolBytes) < 1024 {
+			chunk := make([]byte, 1024)
+			copy(chunk, iVolBytes)
+			audioChunks = append(audioChunks, chunk)
+			break
+		}
+		audioChunks = append(audioChunks, iVolBytes[:1024])
+		iVolBytes = iVolBytes[1024:]
 	}
 
 	return audioChunks
@@ -73,7 +73,7 @@ func lowPassFilter(data []byte, cutoffFreq float64, sampleRate int) []byte {
 	return int16sToBytes(filtered)
 }
 
-// technically copied too
+// copied too
 func downsample24kTo16kLinear(input []byte) []byte {
 	int16s := bytesToInt16s(input)
 	outputLength := (len(int16s) * 2) / 3
@@ -90,19 +90,3 @@ func downsample24kTo16kLinear(input []byte) []byte {
 
 	return int16sToBytes(output)
 }
-
-// entry-point (soxr)
-func downsampleAudioSoxr(input []byte) [][]byte {
-	newBytes := new(bytes.Buffer)
-	dec, _ := resample.New(newBytes, 24000, 16000, 1, resample.I16, resample.VeryHighQ)
-	dec.Write(input)
-	var audioChunks [][]byte
-	decodedBytes := newBytes.Bytes()
-	filteredBytes := lowPassFilter(decodedBytes, 4000, 16000)
-	iVolBytes := increaseVolume(filteredBytes, 5)
-	for len(iVolBytes) >= 1024 {
-		audioChunks = append(audioChunks, iVolBytes[:1024])
-		iVolBytes = iVolBytes[1024:]
-	}
-	return audioChunks
-}

chipper/pkg/wirepod/ttr/kgsim_cmds.go

@@ -104,7 +104,7 @@ type LLMCommand struct {
 var ValidLLMCommands []LLMCommand = []LLMCommand{
 	{
 		Command:         "playAnimationWI",
-		Description:     "Plays an animation on the robot without interrupting speech. This should be used FAR more than the playAnimation command. This is great for storytelling and making any normal response animated. Don't put two of these right next to each other. Use this MANY times. The param choices are the onlyy choices you have. You can't create any.",
+		Description:     "Plays an animation on the robot without interrupting speech. This should be used FAR more than the playAnimation command. This is great for storytelling and making any normal response animated. Don't put two of these right next to each other. Use this MANY times. The param choices are the only choices you have. You can't create any.",
 		ParamChoices:    "happy, veryHappy, sad, verySad, angry, frustrated, dartingEyes, confused, thinking, celebrate, love",
 		Action:          ActionPlayAnimationWI,
 		SupportedModels: []string{"all"},
@@ -307,6 +307,9 @@ func getOpenAIVoice(voice string) openai.SpeechVoice {
 
 // TODO
 func DoSayText_OpenAI(robot *vector.Vector, input string) error {
+	if strings.TrimSpace(input) == "" {
+		return nil
+	}
 	openaiVoice := getOpenAIVoice(vars.APIConfig.Knowledge.OpenAIPrompt)
 	// if vars.APIConfig.Knowledge.OpenAIVoice == "" {
 	// 	openaiVoice = openai.VoiceFable
@@ -318,7 +321,7 @@ func DoSayText_OpenAI(robot *vector.Vector, input string) error {
 		Model:          openai.TTSModel1,
 		Input:          input,
 		Voice:          openaiVoice,
-		ResponseFormat: "pcm",
+		ResponseFormat: openai.SpeechResponseFormatPcm,
 	})
 	if err != nil {
 		logger.Println(err)
@@ -338,12 +341,8 @@ func DoSayText_OpenAI(robot *vector.Vector, input string) error {
 		},
 	})
 	//time.Sleep(time.Millisecond * 30)
-	var audioChunks [][]byte
-	if os.Getenv("USE_GO_DESAMPLE") == "true" {
-		audioChunks = downsample24kTo16k(speechBytes)
-	} else {
-		audioChunks = downsampleAudioSoxr(speechBytes)
-	}
+	audioChunks := downsample24kTo16k(speechBytes)
+
 	var chunksToDetermineLength []byte
 	for _, chunk := range audioChunks {
 		chunksToDetermineLength = append(chunksToDetermineLength, chunk...)
@@ -366,6 +365,8 @@ func DoSayText_OpenAI(robot *vector.Vector, input string) error {
 			},
 		})
 	}()
+	os.WriteFile("../../pre-process.pcm", speechBytes, 0777)
+	os.WriteFile("../../post-process.pcm", chunksToDetermineLength, 0777)
 	time.Sleep(pcmLength(chunksToDetermineLength) + (time.Millisecond * 50))
 	return nil
 }

chipper/webroot/js/main.js

@@ -347,6 +347,7 @@ function updateKGAPI() {
       if (data.provider === "openai") {
         getE("openaiKey").value = data.key;
         getE("openAIPrompt").value = data.openai_prompt;
+        getE("openaiVoice").value = data.openai_voice;
         getE("commandYes").checked = data.commands_enable
         getE("intentyes").checked = data.intentgraph
         getE("saveChatYes").checked = data.save_chat

dockerfile

@@ -3,7 +3,7 @@ FROM ubuntu
 
 COPY . .
 
-RUN chmod +x /setup.sh && apt-get update && apt-get install -y dos2unix && dos2unix /setup.sh && apt-get install -y avahi-daemon avahi-autoipd libsoxr0
+RUN chmod +x /setup.sh && apt-get update && apt-get install -y dos2unix && dos2unix /setup.sh && apt-get install -y avahi-daemon avahi-autoipd
 
 RUN ["/bin/sh", "-c", "STT=vosk ./setup.sh"]
 

setup.sh

@@ -90,13 +90,13 @@ function getPackages() {
     echo "Installing required packages"
     if [[ ${TARGET} == "debian" ]]; then
         apt update -y
-        apt install -y wget openssl net-tools libsox-dev libopus-dev make iproute2 xz-utils libopusfile-dev pkg-config gcc curl g++ unzip avahi-daemon git libasound2-dev libsodium-dev libsoxr-dev
+        apt install -y wget openssl net-tools libsox-dev libopus-dev make iproute2 xz-utils libopusfile-dev pkg-config gcc curl g++ unzip avahi-daemon git libasound2-dev libsodium-dev
         elif [[ ${TARGET} == "arch" ]]; then
         pacman -Sy --noconfirm
-        sudo pacman -S --noconfirm wget openssl net-tools sox opus make iproute2 opusfile curl unzip avahi git libsodium go pkg-config libsoxr 
+        sudo pacman -S --noconfirm wget openssl net-tools sox opus make iproute2 opusfile curl unzip avahi git libsodium go pkg-config 
         elif [[ ${TARGET} == "fedora" ]]; then
         dnf update
-        dnf install -y wget openssl net-tools sox opus make opusfile curl unzip avahi git libsodium-devel soxr-devel
+        dnf install -y wget openssl net-tools sox opus make opusfile curl unzip avahi git libsodium-devel
         elif [[ ${TARGET} == "darwin" ]]; then
         sudo -u $SUDO_USER brew update
         sudo -u $SUDO_USER brew install wget pkg-config opus opusfile