How to Make a Telephone TTS Voice for Menus and Intercoms
Write a clear phone menu, generate a Mike for Telephone source, tune its timing, and export separate WAV clips before adding line effects.

In this guide
Create the voice source before the phone effect
Use Mike for Telephone to generate a short menu prompt, then save the dry WAV. The preset changes SAM synthesis parameters; it does not apply a telephone codec or simulated line noise. Add the channel effect in an editor after the words are clear.
Thank you for calling. Press one for status. Press two to repeat.Speed 86 · Pitch 58 · Mouth 95 · Throat 105
Dry phone-menu source. No telephone band-pass filter or line noise is applied.
Load this exampleStart with the actual telephone preset
The default values are Speed 86, Pitch 58, Mouth 95 and Throat 105. To slow the menu, try Speed 96. To shorten it, try 78. The larger Speed number produces slower speech in this engine.
Keep the other three controls fixed during timing tests. If a station name is unclear, test a phonetic rewrite before modifying resonance. Listen to the complete menu after fixing a word, because sentence context can change pronunciation.
Process a copy of the WAV
Import the dry voice into an audio editor. If the scene needs a narrow line sound, reduce the low and high frequency extremes with a band-pass filter. Preview its effect on consonants rather than treating one filter setting as correct for every voice.
Add any static on a separate track and leave enough level difference for the words to remain intelligible. Put a connection click or beep outside the speech, where it will not mask the first syllable. Retain the clean original for revisions.
Make pauses and interruptions in the destination
For a game menu, export each choice as its own file. The game can then stop playback as soon as the player makes a selection. A single long recording is harder to interrupt or reorder cleanly.
For a video, place deliberate silence between prompts on the timeline. Punctuation in TTS is useful for phrasing, but it is not an exact millisecond pause control. Test the complete sequence on a small speaker before the final export.
