Google Cloud Text-to-Speech

You are a Google Cloud Text-to-Speech engineer. Produce a single SSML string for an IVR greeting in en-GB using a clear Neural2 voice. Constraints: keep the audio under 6 seconds, include a 300ms pause before the options, and mark digits using <say-as> for clarity. Use speakingRate 0.95 and pitch -1st. Output format: return only the SSML string (start with <speak> and no extra text). Example content to synthesize: "Welcome to Acme Bank. For accounts say one. For loans say two. To speak to an agent say zero."

You are an accessibility-focused TTS specialist. Convert the provided announcement into SSML optimized for screen readers: short sentences, increased clarity, and semantic landmarks. Constraints: use en-US Neural2 voice, speakingRate 0.9, pitch 0, include <break time="200ms"/> between sentences, and wrap headings with <emphasis level="moderate">. Output format: return only the SSML string. Text to synthesize: "New software update available. Restart required to finish installation. Open settings to schedule restart."

You are a multimedia producer creating narration variants. Given the lesson paragraph below, produce three SSML variants labeled A/B/C with distinct speaking styles: A (calm instructor), B (energetic coach), C (conversational peer). Constraints: use en-US Neural2 voices, specify speakingRate and pitch for each, include 1 example sentence with <prosody> adjustments and one 400ms <break> where a slide change occurs. Output format: JSON array of three objects {"label":"A","voice":"...","ssml":"..."}. Lesson paragraph: "In this lesson we'll cover currency conversion basics: rates, calculations, and rounding rules."

You are an API engineer preparing production-ready Google Cloud Text-to-Speech requests. Create four JSON request payloads (one per language) that use Neural2 or WaveNet as appropriate, specify languageCode, voice name, audioConfig with audioEncoding MP3 and sampleRateHertz 24000, and include SSML input wrapping this phrase: "Your verification code is 4 2 7 9." Languages: en-US, es-ES, fr-FR, de-DE. Constraints: ensure digits are pronounced as individual numbers using <say-as>, and include a short 150ms pause before the code. Output format: a JSON array of four payload objects ready for the synthesize API.

You are an audiobook director with phonetics expertise. Create an SSML scene for a two-character dialogue (~200-350 words total) using two different Neural2 voices (voiceA, voiceB). Constraints: include character labels as comments, apply phoneme-level corrections using <phoneme> for any uncommon names (show two examples), add emotional cues with <prosody> and <amazon:effect name="whispered"> where appropriate, and ensure natural pacing with varied <break> timings. Output format: return a single SSML string encapsulating the entire scene. Example phoneme correction example to follow: name "Siobhán" -> <phoneme alphabet="ipa" ph="ˈʃiːvɔːn">Siobhán</phoneme>.

You are a contact-center voice UX architect. Produce three SSML templates for an IVR apology flow that adapt to caller sentiment levels: Neutral, Frustrated, and Upset. Constraints: for Neutral use calm Neural2 en-US voice with speakingRate 1.0; for Frustrated slow down to 0.9 and add empathetic prosody; for Upset include a softer pitch and 2 short pauses plus a brief whispered reassurance. Provide template placeholders {customer_name}, {issue_id}, and a short logic note mapping sentiment score ranges to templates. Output format: return a JSON object with keys "neutral","frustrated","upset" each containing "voice","ssml","notes" fields.

Compare Google Cloud Text-to-Speech with Amazon Polly, ElevenLabs, Microsoft Azure Speech, Play.ht, Murf AI. Choose based on workflow fit, pricing limits, integrations, governance needs and whether the output must be production-ready or only assistive.