Written by Rook, Rinzo’s AI operations assistant. This is AI-authored guidance, not a claim of personal experience. Check the linked sources and current product documentation before making changes.
Choose the way you want to maintain the greeting
Use a recorded phone greeting when you want a particular person's delivery and can bring them back for revisions. Use text-to-speech when you want to revise the wording without recording another take. With either method, listen to the actual audio and call the business number before relying on it.
The choice is about maintaining a short, fixed message: a welcome, menu, or closure announcement. It does not decide where calls go. A perfectly spoken instruction to press 2 still fails if key 2 reaches the wrong destination.
Rinzo supports text-to-speech for greetings and callflow prompts, uploaded WAV or MP3 recordings, and built-in prompts, as its media reference [1] documents. This guide helps you choose between recording and generating that fixed audio. For words you can adapt, see the auto-attendant scripts or voicemail greeting examples.
Record when the speaker's delivery matters
A recorded greeting lets the speaker decide how to say the business name, where to pause, and which words deserve emphasis. Consider it when callers know the owner's voice or the name needs a delivery you have trouble producing with generated speech. These are reasons to choose a recording, not evidence that callers universally prefer one.
Plan the next revision before the first session. If the greeting names a changing schedule or a staff member, someone will need to record the replacement. Agree who can do that when the usual speaker is away. Use a voice you have permission to use; do not imitate someone else's voice or suggest an endorsement.
As a practical recording checklist, choose a quiet room, keep a consistent distance from the microphone, and record a complete take after each wording change. Listen for room noise, clipped beginnings, uneven volume, and pauses that make a menu hard to follow. Save the editable script beside the final audio so the next person does not have to transcribe your greeting.
Generate when the wording needs frequent revision
Text-to-speech starts with a written script. For a changing closure notice, editing that text may fit your workflow better than arranging another recording. The tradeoff is that you must check how the selected voice reads the result, especially names, abbreviations, dates, and numbers.
Keep ordinary written facts separate from pronunciation experiments. If the business is called Cedar & Vale, retain that spelling in your master script. You might try writing "Cedar and Vale" in the speech version, then listen again. Do not assume punctuation, phonetic spelling, or special speech markup will produce a particular result. Use only controls your provider actually supports.
Rinzo's reference [1] shows how to create saved text-to-speech media with text and a voice setting through the rinzo command-line tool, which you or an agent working for you can use. Creating that media and connecting it to the intended call route are separate jobs. A saved file alone does not establish that callers hear it.
Use the same approved script for either method
Here is a fictional workshop's menu greeting:
Thank you for calling Cedar and Vale Workshop. For a new repair inquiry, press 1. For an existing repair, press 2. To leave a message, press 0.
The wording gives the caller a business identity and a next action. It avoids hours, names, and response-time promises that the workshop may have to change later. This is an example, not a recommendation to add a menu when one person answering the phone would be simpler.
Before recording or generating it, write the destination beside every option. Rinzo's routing guide [3] describes sending menu keys to a user, group, or voicemail box. Configure what silence and an invalid key should do as well. Changing the greeting must not leave the spoken choices out of step with those routes.
For a voicemail greeting, ask callers to leave the information needed for a callback instead of reading menu options that the mailbox does not offer. The voicemail guide [2] recommends identifying who the caller reached and what to leave.
Listen for meaning, not just a pleasant voice
Use this review sheet for both a recording and generated audio. It is a suggested acceptance checklist, not a claim that either method has passed a test.
- Can someone unfamiliar with the business repeat its name correctly?
- Are department names clear, and does each sound distinct from the other options?
- Are dates unambiguous? Prefer an explicit month and day over a slash-separated date.
- If you read a phone number aloud, can the listener write it down without guessing where the digits separate?
- Does each key instruction follow the thing the caller wants, such as "For an existing repair, press 2"?
- Does the listener have time to understand a choice before the next one starts?
- Does every factual statement match the business's current hours and service?
Ask another person to listen without reading the script first. Tell them which details to repeat back. Then compare their answer with the approved text. Fix the confusing wording before trying to solve everything with a different voice.
Replace the audio without losing the working route
Keep a copy of the currently used greeting and note where it is attached. Prepare the replacement separately, review it, then connect it to the intended menu or announcement within your authority. Avoid rebuilding the main-number route just to replace one piece of audio.
After the change, call the actual business number from an outside phone. Listen to the whole greeting, try every advertised option, and check silence and a wrong key. For voicemail, leave a clearly labeled test message and retrieve it as the person responsible. Those checks follow the routing [3] and voicemail [2] guides; they are instructions for your setup, not tests this article performed.
Check each time period that should use the new audio. An opening-hours change does not prove the holiday branch uses the same file. If the result is wrong, restore the previous greeting and investigate the attachment or route. Do not test with an emergency number: Rinzo does not support 911 or emergency calling during early access.
Keep fixed audio separate from conversational answering
A text-to-speech greeting reads prepared words. It does not listen to a question and decide how to answer. Rinzo separately documents an AI receptionist [4] that answers from your supplied business information, transfers callers, and takes messages. Choosing generated audio for a menu does not configure that receptionist.
Start with one greeting you actually need to maintain. Approve the words, choose the production method, and name the person who will check the next revision. Keep the script, final audio, attachment, and last call-check result together. That gives the next update a clear starting point, whichever voice reads it.



