AI Studio · Service

    One script.
    Every voice.

    A voice chosen for the brand, then used for ads, product videos, IVR lines and podcast segments, in Hindi, English and other languages. Generated from your script, checked by ear, mastered by an audio editor.

    See pricing
    At a glance
    From $0 per minute
    Standard delivery: 24 hours once the voice is settled
    0 hr
    Standard delivery
    0+
    Projects since 2019
    WHAT'S INCLUDED

    Every read. One voice.

    The voice is settled once. Every script after it uses the same settings.

    The voice
    01
    Voice profile: tone, pace, age and accent, written down
    02
    A shortlist of voices, each reading your own script
    03
    A licensed library voice, or a clone of a speaker who has consented in writing (a clone is quoted separately)
    04
    A pronunciation list for brand, product and ingredient names
    05
    The same voice carried across languages, where the tools support it (each language quoted)
    The audio
    01
    Voiceover for ads, product videos and tutorials
    02
    Podcast intro and outro, with music (quoted)
    03
    Ad reads for podcast sponsorships and radio
    04
    IVR greetings, menus, on-hold messages and notification sounds (quoted by count)
    05
    Versions in Hindi, English, Tamil, Marathi and other supported languages
    Editing and handover
    01
    Editing and mastering in Adobe Audition
    02
    WAV masters, and MP3 for the web
    03
    IVR files in the format your telephony provider asks for
    04
    Commercial rights to every file delivered
    05
    The voice settings and pronunciation list, so later scripts match
    WHAT IT COSTS

    Priced by the minute.

    $69 per finished minute of audio: one language, a licensed library voice chosen from three samples, a pronunciation list, WAV and MP3, and two rounds of changes. A cloned voice, extra languages, IVR and podcast work are quoted in writing.

    $0
    Per minute of audio
    One language, a library voice, two rounds of changes. Anything beyond is quoted within 48 hours.
    0 hr
    Standard delivery
    For a script in a voice already settled. Choosing the voice comes first.
    0 min
    First call
    The scripts, the languages and where the audio will play, then a written scope.
    HOW LONG

    The voice, once. Then scripts.

    The first two happen once. The last two repeat with every script.

    01
    Day 1–3

    Voice profile

    Tone, pace and character of the voice agreed in writing, with the languages and where the audio will be used.

    02
    Day 3–7

    Samples and sign-off

    A shortlist of voices reading your script, inside 48 hours. One is chosen, or a consenting speaker is recorded for a clone.

    03
    Per script

    Reads

    Each script generated in the settled voice. A person listens to every file and fixes pronunciation, pacing and emphasis.

    04
    24 hours from script

    Master and deliver

    Edited, mastered to the loudness the channel needs, and exported in every format.

    IF IT GOES WRONG

    Samples first. In your words.

    You hear the voice reading your own script before anything is produced.

    01

    Samples before production

    A shortlist reads your script. Production starts after you choose.

    02

    Consent in writing for a cloned voice

    A real person's voice is cloned only with that person's written consent. Celebrity and sound-alike voices are declined.

    03

    A person listens to every file

    Mispronounced names, odd stresses and clipped words are caught by ear and regenerated before delivery.

    04

    A changed line is a regeneration

    Script edits are regenerated in the same voice and re-mastered. The scope says how many rounds are included.

    05

    Commercial rights

    Every delivered file comes with the right to use it in ads, videos, IVR and podcasts.

    06

    Scope in writing

    Scripts, languages, formats and revision rounds are listed. A change is priced before it is made.

    WHO IT'S FOR

    Scripts, yes. Acting, no.

    If the second list sounds like you, a voice artist or a neighbouring service is the better choice.

    Generated voice fits

    • 01
      Product videos in several languages

      One script, voiced for each market in the same voice.

    • 02
      Audio that changes often

      Offers, prices and IVR menus rewritten every month.

    • 03
      A podcast that needs a frame

      Intro, outro, sponsor reads and mastering around the conversation you record yourself.

    • 04
      Customer service lines

      IVR greetings, menus and on-hold messages in one steady voice.

    Look elsewhere first

    • 01
      The read is a performance

      Character acting, singing, comedy timing or an emotional brand film. Hire a voice artist.

    • 02
      The founder's own voice matters

      A founder's message should be the founder, recorded. Editing and mastering of that recording is part of this service.

    • 03
      The audio needs pictures

      A finished video with voiceover sits in AI Video & UGC.

    STANDARDS YOU CAN CHECK

    Mastered to a number. Checked by ear.

    Each one can be checked on the delivered files.

    01

    Masters

    WAV at 48 kHz, 24-bit. MP3 alongside for the web.

    02

    Loudness

    Podcasts at -16 LUFS, social and streaming at -14 LUFS, true peak under -1 dB.

    03

    IVR

    Mono at 8 kHz, in the codec your telephony provider specifies.

    04

    Pronunciation

    Brand, product and ingredient names follow the agreed list in every file and every language.

    05

    Clean edits

    No clipped breaths, doubled words or jumps in tone between regenerated lines.

    06

    File naming

    Project, language and version in every file name, with the script attached as text.

    BEFORE YOU BRIEF

    Before the first read.

    It depends on the route. A voice cloned from a speaker you bring, with their written consent, is used for your brand only. A voice chosen from a tool's library is licensed for your commercial use but is not exclusive: other users of the same tool can pick it. The scope says which route you are on.
    Yes, with that person's written consent, and the cloning tools ask the speaker to verify it themselves. A voice is never cloned from someone who has not agreed, and celebrity or sound-alike voices are declined.
    Natural enough for ads, product videos and IVR. Long emotional reads are where a generated voice shows. You hear samples reading your own script before anything is produced.
    Hindi, English, Tamil, Marathi and the other languages the voice tools support, ten or more in all. A pronunciation list covers brand, product and ingredient names. For a language nobody on the project speaks, a sample goes to a speaker you trust before the batch is run.
    Yes. Every delivered file comes with commercial rights, written into the scope: ads, videos, IVR and podcasts. Music in intros and outros is cleared for the same use. Where a platform asks for a label on realistic generated voices, the delivery note says so.
    Once the voice is settled, a standard script comes back within 24 hours, generated, checked by ear and mastered. Choosing the voice comes first and takes a few days, with samples inside 48 hours.
    The frame around your conversation: intro and outro with music, sponsor reads, and editing and mastering of the episodes you record. The conversation itself should be you.
    The changed lines are regenerated in the same voice and re-mastered into the file. Nobody is re-booked. The scope says how many rounds of changes are included.
    For a performance: character acting, singing, comedy timing, or the voice of an emotional brand film. A founder's message should be the founder, recorded. Generated voice is for scripted reads that change often or run in many languages.

    Your script.
    Out loud.

    A 30-minute call about the scripts and the languages. A written scope after it.

    See pricing