improved voicebot example

examples/local-voicebot/main.go

@@ -26,7 +26,7 @@ func main() {
 
 	chatmodel := "gpt-4o"
 	audiomodel := "whisper-1"
-	speakmodel := "tts-1-hd"
+	speakmodel := "tts-1"
 
 	url := os.Getenv("OPENAI_API_BASE")
 
@@ -43,7 +43,7 @@ func main() {
 	client := openai.NewClient(options...)
 
 	messages := []openai.ChatCompletionMessageParamUnion{
-		openai.SystemMessage("Your knowledge cutoff is 2023-10. You are a helpful, witty, and friendly AI. Act like a human, but remember that you aren't a human and that you can't do human things in the real world. Your voice and personality should be warm and engaging, with a lively and playful tone. If interacting in a non-English language, start by using the standard accent or dialect familiar to the user. Talk quickly. You should always call a function if you can. Answer as briefly and concisely as possible."),
+		openai.SystemMessage("Your knowledge cutoff is 2023-10. You are a helpful, witty, and friendly AI. Act like a human, but remember that you aren't a human and that you can't do human things in the real world. Your voice and personality should be warm and engaging, with a lively and playful tone. If interacting in a non-English language, start by using the standard accent or dialect familiar to the user. Talk quickly. You should always call a function if you can. Answer as briefly and concisely as possible. Keep it short."),
 	}
 
 	for ctx.Err() == nil {
@@ -58,7 +58,7 @@ func main() {
 
 		transcription, err := client.Audio.Transcriptions.New(ctx, openai.AudioTranscriptionNewParams{
 			Model: openai.F(audiomodel),
-			File:  openai.F[io.Reader](bytes.NewReader(data)),
+			File:  openai.FileParam(bytes.NewReader(data), "file.wav", "audio/wav"),
 		})
 
 		if err != nil {
@@ -81,6 +81,8 @@ func main() {
 
 		completion := openai.ChatCompletionAccumulator{}
 
+		print("📣 ")
+
 		for stream.Next() {
 			chunk := stream.Current()
 			completion.AddChunk(chunk)
@@ -90,6 +92,8 @@ func main() {
 			}
 		}
 
+		println()
+
 		if err := stream.Err(); err != nil {
 			println("error:", err.Error())
 			continue
@@ -98,8 +102,6 @@ func main() {
 		message := completion.Choices[0].Message
 		messages = append(messages, message)
 
-		println("📣 " + message.Content)
-
 		sayText(ctx, client, speakmodel, message.Content)
 	}
 }
@@ -154,21 +156,21 @@ func recordChunk(ctx context.Context) ([]byte, error) {
 		args = []string{
 			"-f", "avfoundation",
 			"-i", ":0",
-			"-af", "silencedetect=noise=-30dB:d=2",
+			"-af", "silencedetect=noise=-30dB:d=1",
 			path,
 		}
 	case "windows":
 		args = []string{
 			"-f", "dshow",
 			"-i", "audio=default",
-			"-af", "silencedetect=noise=-30dB:d=2",
+			"-af", "silencedetect=noise=-30dB:d=1",
 			path,
 		}
 	case "linux":
 		args = []string{
 			"-f", "alsa",
 			"-i", "default",
-			"-af", "silencedetect=noise=-30dB:d=2",
+			"-af", "silencedetect=noise=-30dB:d=1",
 			path,
 		}
 	}
@@ -206,7 +208,13 @@ func recordChunk(ctx context.Context) ([]byte, error) {
 		}
 	}
 
-	if err := cmd.Process.Kill(); err != nil {
+	err = cmd.Process.Signal(os.Interrupt)
+
+	if err != nil {
+		err = cmd.Process.Kill()
+	}
+
+	if err != nil {
 		fmt.Println("Error killing FFmpeg process:", err)
 		return nil, err
 	}

pkg/provider/openai/synthesizer.go

@@ -41,7 +41,8 @@ func (s *Synthesizer) Synthesize(ctx context.Context, content string, options *p
 		Model: openai.F(s.model),
 		Input: openai.F(content),
 
-		Voice:          openai.F(openai.AudioSpeechNewParamsVoiceAlloy),
+		Voice: openai.F(openai.AudioSpeechNewParamsVoiceAlloy),
+
 		ResponseFormat: openai.F(openai.AudioSpeechNewParamsResponseFormatWAV),
 	})
 

pkg/provider/openai/transcriber.go

@@ -41,7 +41,7 @@ func (t *Transcriber) Transcribe(ctx context.Context, input provider.File, optio
 
 	transcription, err := t.transcriptions.New(ctx, openai.AudioTranscriptionNewParams{
 		Model: openai.F(t.model),
-		File:  openai.F(input.Content),
+		File:  openai.FileParam(input.Content, input.Name, ""),
 	})
 
 	if err != nil {