feat: Add missing TTS models and voices (#958)

* feat: Add missing TTS models and voices

* feat: Add new instruction field to create speech request

- From docs: Control the voice of your generated audio with additional instructions. Does not work with tts-1 or tts-1-hd.

* fix: add canary-tts back to SpeechModel

Author: netr

speech.go

@@ -8,20 +8,25 @@ import (
 type SpeechModel string
 
 const (
-	TTSModel1      SpeechModel = "tts-1"
-	TTSModel1HD    SpeechModel = "tts-1-hd"
-	TTSModelCanary SpeechModel = "canary-tts"
+	TTSModel1         SpeechModel = "tts-1"
+	TTSModel1HD       SpeechModel = "tts-1-hd"
+	TTSModelCanary    SpeechModel = "canary-tts"
+	TTSModelGPT4oMini SpeechModel = "gpt-4o-mini-tts"
 )
 
 type SpeechVoice string
 
 const (
 	VoiceAlloy   SpeechVoice = "alloy"
+	VoiceAsh     SpeechVoice = "ash"
+	VoiceBallad  SpeechVoice = "ballad"
+	VoiceCoral   SpeechVoice = "coral"
 	VoiceEcho    SpeechVoice = "echo"
 	VoiceFable   SpeechVoice = "fable"
 	VoiceOnyx    SpeechVoice = "onyx"
 	VoiceNova    SpeechVoice = "nova"
 	VoiceShimmer SpeechVoice = "shimmer"
+	VoiceVerse   SpeechVoice = "verse"
 )
 
 type SpeechResponseFormat string
@@ -39,6 +44,7 @@ type CreateSpeechRequest struct {
 	Model          SpeechModel          `json:"model"`
 	Input          string               `json:"input"`
 	Voice          SpeechVoice          `json:"voice"`
+	Instructions   string               `json:"instructions,omitempty"`    // Optional, Doesnt work with tts-1 or tts-1-hd.
 	ResponseFormat SpeechResponseFormat `json:"response_format,omitempty"` // Optional, default to mp3
 	Speed          float64              `json:"speed,omitempty"`           // Optional, default to 1.0
 }