diff options
| author | Paul Buetow <paul@buetow.org> | 2026-04-01 20:20:04 +0300 |
|---|---|---|
| committer | Paul Buetow <paul@buetow.org> | 2026-04-01 20:20:04 +0300 |
| commit | 4f1d809b83c373aa27cd453bd7b3e04ab9719878 (patch) | |
| tree | da49fd3c9a73079eaf41d8fcefee24d16b85e8dc | |
| parent | 4727a188102d8166e4acabc4fd62863245e6cdce (diff) | |
zg: extract shared audio voice lists
| -rw-r--r-- | internal/audio/provider.go | 4 | ||||
| -rw-r--r-- | internal/audio/voices.go | 46 | ||||
| -rw-r--r-- | internal/audio/voices_test.go | 36 | ||||
| -rw-r--r-- | internal/gui/generator.go | 14 | ||||
| -rw-r--r-- | internal/processor/processor.go | 10 |
5 files changed, 91 insertions, 19 deletions
diff --git a/internal/audio/provider.go b/internal/audio/provider.go index b961482..334e709 100644 --- a/internal/audio/provider.go +++ b/internal/audio/provider.go @@ -26,14 +26,14 @@ type Config struct { // OpenAI-specific settings OpenAIKey string OpenAIModel string // "tts-1", "tts-1-hd", or "gpt-4o-mini-tts" - OpenAIVoice string // "alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse" + OpenAIVoice string // One of OpenAIVoices. OpenAISpeed float64 // 0.25 to 4.0 OpenAIInstruction string // Voice instructions for gpt-4o-mini-tts model // Gemini-specific settings GoogleAPIKey string GeminiTTSModel string // "gemini-2.5-flash-preview-tts" - GeminiVoice string // Prebuilt Gemini TTS voice name, or empty for the model default + GeminiVoice string // One of GeminiVoices, or empty for the model default. GeminiSpeed float64 // Prompt hint for desired speech speed } diff --git a/internal/audio/voices.go b/internal/audio/voices.go new file mode 100644 index 0000000..9a96c76 --- /dev/null +++ b/internal/audio/voices.go @@ -0,0 +1,46 @@ +package audio + +// OpenAIVoices lists the OpenAI voices supported by the app. +var OpenAIVoices = []string{ + "alloy", + "ash", + "ballad", + "coral", + "echo", + "fable", + "onyx", + "nova", + "sage", + "shimmer", + "verse", +} + +// GeminiVoices lists the Gemini prebuilt voices supported by the app. +var GeminiVoices = []string{ + "Zephyr", + "Puck", + "Charon", + "Kore", + "Fenrir", + "Leda", + "Orus", + "Aoede", + "Callirrhoe", + "Autonoe", + "Enceladus", + "Iapetus", + "Umbriel", + "Algieba", + "Despina", + "Erinome", + "Gacrux", + "Pulcherrima", + "Achernar", + "Rasalgethi", + "Laomedeia", + "Sadachbia", + "Schedar", + "Sulafat", + "Vindemiatrix", + "Zubenelgenubi", +} diff --git a/internal/audio/voices_test.go b/internal/audio/voices_test.go new file mode 100644 index 0000000..121f328 --- /dev/null +++ b/internal/audio/voices_test.go @@ -0,0 +1,36 @@ +package audio + +import ( + "reflect" + "testing" +) + +func TestVoiceLists(t *testing.T) { + t.Parallel() + + tests := []struct { + name string + got []string + want []string + }{ + { + name: "openai", + got: OpenAIVoices, + want: []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"}, + }, + { + name: "gemini", + got: GeminiVoices, + want: []string{"Zephyr", "Puck", "Charon", "Kore", "Fenrir", "Leda", "Orus", "Aoede", "Callirrhoe", "Autonoe", "Enceladus", "Iapetus", "Umbriel", "Algieba", "Despina", "Erinome", "Gacrux", "Pulcherrima", "Achernar", "Rasalgethi", "Laomedeia", "Sadachbia", "Schedar", "Sulafat", "Vindemiatrix", "Zubenelgenubi"}, + }, + } + + for _, tt := range tests { + t.Run(tt.name, func(t *testing.T) { + t.Parallel() + if !reflect.DeepEqual(tt.got, tt.want) { + t.Fatalf("%s voice list mismatch\nwant: %#v\ngot: %#v", tt.name, tt.want, tt.got) + } + }) + } +} diff --git a/internal/gui/generator.go b/internal/gui/generator.go index 7964e72..3491db2 100644 --- a/internal/gui/generator.go +++ b/internal/gui/generator.go @@ -63,11 +63,8 @@ func (a *Application) generateAudio(ctx context.Context, word string, cardDir st } } - // Always use random voice and speed - allVoices := []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"} - // Select a random voice - voice, speed := randomVoiceAndSpeed(allVoices) + voice, speed := randomVoiceAndSpeed(audio.OpenAIVoices) // Create a copy of audio config with selected voice and speed audioConfig := *a.audioConfig @@ -127,8 +124,7 @@ func (a *Application) generateAudioFront(ctx context.Context, word string, cardD return "", fmt.Errorf("card directory not provided") } - allVoices := []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"} - voice, speed := randomVoiceAndSpeed(allVoices) + voice, speed := randomVoiceAndSpeed(audio.OpenAIVoices) audioConfig := *a.audioConfig audioConfig.OpenAIVoice = voice @@ -169,8 +165,7 @@ func (a *Application) generateAudioBack(ctx context.Context, text string, cardDi return "", fmt.Errorf("card directory not provided") } - allVoices := []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"} - voice, speed := randomVoiceAndSpeed(allVoices) + voice, speed := randomVoiceAndSpeed(audio.OpenAIVoices) audioConfig := *a.audioConfig audioConfig.OpenAIVoice = voice @@ -201,8 +196,7 @@ func (a *Application) generateAudioBgBg(ctx context.Context, front, back, cardDi return "", "", fmt.Errorf("card directory not provided") } - allVoices := []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"} - voice, speed := randomVoiceAndSpeed(allVoices) + voice, speed := randomVoiceAndSpeed(audio.OpenAIVoices) audioConfig := *a.audioConfig audioConfig.OpenAIVoice = voice diff --git a/internal/processor/processor.go b/internal/processor/processor.go index d7025a6..d2a380b 100644 --- a/internal/processor/processor.go +++ b/internal/processor/processor.go @@ -241,19 +241,17 @@ func (p *Processor) ProcessWordWithTranslationAndType(word, providedTranslation // generateAudio generates audio files for a word func (p *Processor) generateAudio(word string) error { - allVoicesList := []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"} - // Get list of voices to use var voices []string if p.flags.AllVoices { - voices = allVoicesList + voices = audio.OpenAIVoices } else if p.flags.OpenAIVoice != "" { // Use explicitly specified voice voices = []string{p.flags.OpenAIVoice} fmt.Printf(" Using specified voice: %s\n", p.flags.OpenAIVoice) } else { // Select a random voice - randomVoice := allVoicesList[rand.Intn(len(allVoicesList))] + randomVoice := audio.OpenAIVoices[rand.Intn(len(audio.OpenAIVoices))] voices = []string{randomVoice} fmt.Printf(" Using random voice: %s\n", randomVoice) } @@ -273,10 +271,8 @@ func (p *Processor) generateAudio(word string) error { // generateAudioBgBg generates audio files for both sides of a bg-bg card func (p *Processor) generateAudioBgBg(front, back string) error { - allVoicesList := []string{"alloy", "ash", "ballad", "coral", "echo", "fable", "onyx", "nova", "sage", "shimmer", "verse"} - // Select a random voice (same voice for both sides for consistency) - voice := allVoicesList[rand.Intn(len(allVoicesList))] + voice := audio.OpenAIVoices[rand.Intn(len(audio.OpenAIVoices))] if p.flags.OpenAIVoice != "" { voice = p.flags.OpenAIVoice } |
