hear every caller voice

each sample is the same line, recorded directly from the voice model at phone quality. these samples let you compare delivery. our received call tests check what survives a carrier SIP route. every voice speaks the five most spoken languages: english, mandarin chinese, hindi, spanish, arabic.

read the line in each language

english: "Hi! This is Alex's assistant. I was hoping to get a table for four tonight, around seven. Do you have anything open?"

mandarin chinese: "您好!我是Alex的助理。我想订今晚七点左右四个人的位子,请问还有空位吗?" (Hello! I'm Alex's assistant. I'd like to book a table for four tonight around seven. Is there still space?)

hindi: "नमस्ते! Alex की तरफ़ से कॉल है। आज रात करीब सात बजे चार लोगों के लिए टेबल मिल सकती है क्या?" (Hello! Calling for Alex. Could we get a table for four people tonight around seven?)

spanish: "¡Hola! Llamo de parte de Alex. Quería reservar una mesa para cuatro esta noche, alrededor de las siete. ¿Tienen algo disponible?" (Hi! I'm calling for Alex. I wanted to book a table for four tonight, around seven. Do you have anything available?)

arabic: "مرحباً! أتصل من طرف أليكس. أودّ حجز طاولة لأربعة أشخاص الليلة حوالي الساعة السابعة. هل لديكم شيء متاح؟" (Hello! I'm calling for Alex. I'd like to book a table for four tonight around seven. Do you have anything available?)

calls in other languages

tell your agent which language to start in ("speak french first"). on real calls to montreal businesses, the caller held the whole call in french, and when one business answered in french without being asked, the caller replied in french on its own.

pick a voice

tell your agent which voice to use when it places the call ("use the gleam voice"). it passes it to the call tool:

call4me_place_call({ ..., voice: "gleam" })

no voice means marin. every voice introduces itself as your assistant, with no name of its own. if you want one ("hi, I'm Sam, Alex's assistant"), ask your agent to save assistant_name in your profile.

why the voices work this way / listen to real calls / install mcp