Function synthesizeWithGoogle

apps/sim/app/api/tools/tts/unified/route.ts:590–693

(
  params: Partial<GoogleTtsParams>
)

}

async function synthesizeWithGoogle(
  params: Partial<GoogleTtsParams>
): Promise<{ audioBuffer: Buffer; format: string; mimeType: string }> {
  const {
    text,
    apiKey,
    voiceId,
    languageCode,
    gender,
    audioEncoding = 'MP3',
    speakingRate = 1.0,
    pitch = 0.0,
    volumeGainDb,
    sampleRateHertz,
    effectsProfileId,
  } = params

  if (!text || !apiKey || !languageCode) {
    throw new Error('text, apiKey, and languageCode are required for Google Cloud TTS')
  }

  const clampedSpeakingRate = Math.max(0.25, Math.min(2.0, speakingRate))

  const audioConfig: Record<string, unknown> = {
    audioEncoding,
    speakingRate: clampedSpeakingRate,
    pitch,
  }

  if (volumeGainDb !== undefined) {
    audioConfig.volumeGainDb = volumeGainDb
  }
  if (sampleRateHertz) {
    audioConfig.sampleRateHertz = sampleRateHertz
  }
  if (effectsProfileId && effectsProfileId.length > 0) {
    audioConfig.effectsProfileId = effectsProfileId
  }

  // Build voice config based on what's provided
  const voice: Record<string, unknown> = {
    languageCode,
  }

  // If voiceId is provided, use it (it takes precedence over gender)
  if (voiceId) {
    voice.name = voiceId
  }

  // Only include gender if specified (don't default to NEUTRAL as it's not supported)
  if (gender) {
    voice.ssmlGender = gender
  }

  // If neither voiceId nor gender is provided, default to a specific voice
  if (!voiceId && !gender) {
    voice.name = 'en-US-Neural2-C'
  }