Telegram

How to make Hindi voiceovers inside Telegram: pick a voice, send a long script as a file, post the audio to your channel

TTS Maker10 min read

हिन्दी में पढ़ें

Your channel is on Telegram, your scripts sit in a chat there, and you would like the voiceover made in the same app. The TTS Maker bot does that: you send text and an MP3 comes back. This guide starts after that first audio. It covers where the Hindi voices are in the bot's menus, what to do when a script is longer than one message, how the bot reads a TXT file, and how to get the audio into your channel without the bot's caption on it.

What the bot does with what you send

Open the bot from the Telegram bot page and press Start. That creates a free account with 5,000 welcome credits, set to the free Microsoft engine and the Hindi voice Madhur. The phone guide walks through the first message. The menus are in English. What comes back depends on what you send.

You sendThe bot
A text messageTurns it into one MP3. A message can hold up to 4,096 characters, the figure on the bot's own Generate screen.
A TXT, PDF, DOCX or EPUB fileShows the character count, the estimated length and the cost, then waits for you to press Create Audiobook.
A message that starts with /Reads it as a command and never speaks it. Do not begin a message with a slash.
A voice note, audio, photo, video or stickerAsks for text or a book file. It does not turn speech into text or read the words in a picture.
Text, straight after you tapped Search voice, Custom Voice ID, Set Style Prompt or Redeem a keyTakes it as the answer it was waiting for, not as a script. Send /cancel first.

Where the Hindi voices are

Send /voices, or tap Voice under any audio made from a message. What opens depends on the engine your account is on, and each engine keeps its Hindi voices in a different place.

EngineHindi voicesHow to reach themCost
Microsoft Edge (Free)2: Madhur (male) and Swara (female)/voices while you are on this engine, then Hindi (India), the first button in the language listFree, up to 200,000 characters a day
ElevenLabs101/engine, then ElevenLabs. The first list has 41 voices and no language filter. Tap Voice Library (494 voices) below it, then Hindi (101)1 credit per character
Gemini AI30/engine, then Gemini AI, then Hindi (30) on the language screen2 credits per character
Cappy VoicesNoneIts list covers eight languages, and Hindi is not one of them1 credit per character

The ElevenLabs Hindi list runs to six pages. The ♂ Male and ♀ Female buttons shorten it, and Search voice finds a voice by name. We cannot hear the voices for you, and the clip the bot sends when you pick one is not your script, so send the same two Hindi lines to each voice you are considering.

The 5,000 welcome credits cover 5,000 characters on an ElevenLabs voice, about five and a half minutes at roughly 900 characters a minute, or 2,500 characters on Gemini AI. /credits shows your balance, the free characters left today and each engine's rate. Plans are on the pricing page.

When the script is longer than one message

4,096 characters is about four and a half minutes of speech. An eight-minute video needs around 7,200. We are not Telegram, so we will not say what your app does with a paste that is too long for one message. What the bot does is in its code: it handles every message on its own. A script that reaches it as two messages comes back as two audios, or as one audio and a reply such as “Please wait — your previous request is still being generated”. It never comes back as one file. Past 4,096 characters, save the script as a .txt file and send the file.

ScriptCharacters, at about 900 a minuteSend it as
A 30-second Shortabout 450One message
Four and a half minutes4,096One message, the most it can hold
An 8-minute videoabout 7,200A TXT file
A 22-minute episodeabout 20,000A TXT file
About 3 hours 40 minutes200,000A TXT file. This is the whole free allowance for one day

What the bot does with a TXT file

  1. It shows a summary first: the file name, the character count, the estimated audio length, the voice and engine, and either FREE or the cost in credits beside your balance. Nothing is made or charged yet.
  2. Check the voice line. If it is the wrong voice, press Cancel, change the voice and send the file again.
  3. Press Create Audiobook. A progress message counts parts: the bot cuts a long text at the ends of sentences and joins the audio afterwards.
  4. The MP3 arrives in the chat, captioned with the name of your file. A very long file comes as several MP3s, marked Part 1, Part 2 and so on. If the conversion fails, the credits or free characters it took are returned.

Five details decide whether it comes out right:

  • Limits. A file can be up to 20 MB, the most a bot can download according to Telegram's Bots FAQ, and 1,000,000 characters. On a free voice the whole file must also fit into the free characters you have left today, out of 200,000. If it does not, the bot answers “Daily free limit reached”.
  • Save the text as UTF-8. That is the encoding the bot tries first. If your notes or file app asks which one to use, choose UTF-8, or the Hindi can come out as the wrong letters.
  • End every line with a full stop or a पूर्ण विराम. The bot may join a line that does not end in . ? ! or । to the next line, unless a blank line sits between them. That repairs text copied from a PDF, where lines are cut short, but a heading with no punctuation can run into the sentence below it. Punctuate headings and leave a blank line between sections.
  • Name the file in English letters. The MP3 takes the file's name: episode-12.txt comes back as episode-12.mp3. From a Devanagari name the matras are dropped, so कहानी भाग 2.txt gives कहन भग 2.mp3. The caption keeps the full name.
  • There is no preview of the text. For a PDF you have not checked, the Audiobook page on the website shows the first lines it extracted before you start. The audiobook guide covers scanned PDFs and old Hindi fonts.
A TXT file laid out for the bot, with one blank line between the parts

अध्याय एक। नई साइकिल।

मीना को जन्मदिन पर एक लाल साइकिल मिली। वह उसे लेकर सीधे मैदान की ओर भागी।

अध्याय दो। पहली बार गिरना।

मोड़ पर साइकिल फिसली, और मीना घुटनों के बल गिर पड़ी। उसने धूल झाड़ी, और फिर से पैडल पर पैर रखा।

Getting the audio into your channel or group

The bot is built for a one-to-one chat. It posts nothing to a channel and does not read channel posts, and nothing in it is made for groups, so do not rely on it there. Make the audio in your own chat with the bot, then move it.

  • Know what you are moving. A message comes back as an audio file, which Telegram's Bot API describes as treated as music by its apps, with a title and a performer. The bot sets the title to the first six words of your text and the performer to the voice name. The caption under it lists the voice, the engine, the character count, the seconds taken, and the word “free” or, on a paid voice, the credits used and your remaining balance.
  • Forward it without the bot's name and caption. Telegram's announcement of its forwarding options, from August 2021, says to tap the “Forward Message” label above the message bar to open a preview, where you can hide the sender's name or hide captions on media messages. Menus change, so look for these in your own app.
  • Write the first six words on purpose. The title belongs to the audio, not to the caption, so expect it to stay when the caption is hidden. The sample below would be titled “नमस्ते दोस्तो! आज की ऑडियो पोस्ट”.
  • Or save the file and upload it yourself. Every audio made from a message is named voice.mp3, so rename it as you save it.
  • Never forward the /web message. Its button opens the website already logged in to your account, and the bot's own message says it works like a password for 30 days.

Try it here

A short audio post for a quiz channel, in Hindi. Press Convert to speech to hear the script, then put in your own day and time.

Your generated audio will appear here.

Commands worth remembering

Every screen can also be reached with the buttons under the bot's messages. The full list of commands is on the Telegram bot page.

CommandWhat it does
/voicesOpens the voice list of the engine you are on.
/engineSwitches engine. Your voice becomes the new engine's first voice, so pick one straight afterwards.
/creditsShows your balance, the free characters left today and each engine's rate.
/historyLists your last ten conversions: date and time, characters, engine and cost.
/cancelClears whatever the bot is waiting for, so your next message becomes audio again.

For speed and pitch, tap Settings under any audio made from a message. On the Microsoft voices, Speed moves in steps of 10 between −50% and +100%, and Pitch in steps of 5 between −50 Hz and +50 Hz. Both are saved to your account.

When the bot is the wrong tool

  • You want a voice message. The bot sends every conversion as an audio file, not as a voice message.
  • You want it to post to your channel by itself. It does not, and nothing in it is made for groups.
  • You are still editing a script of 4,097 to 20,000 characters. Studio on the website takes 20,000 characters in one conversion, so you can change a word and convert again without saving a new file each time.

Questions people ask

Is the Telegram bot free for Hindi text to speech?

Yes, on the two Microsoft Hindi voices, Madhur and Swara, for up to 200,000 characters a day. That allowance is shared with the website, because both run on one account. The 101 ElevenLabs and 30 Gemini AI Hindi voices use credits, and a new account starts with 5,000.

My script is longer than 4,096 characters. How do I send it?

Save it as a .txt file and send the file to the bot. It shows the character count, the estimated length and the cost, then makes the MP3 after you press Create Audiobook. A file can be up to 20 MB and 1,000,000 characters.

Can I add the bot to my channel or group so that it posts the audio there?

No. The bot does not post to channels or read them, and nothing in it is made for groups. It is built for a one-to-one chat. Make the audio there, then forward it or upload the MP3 to your channel yourself.

I sent text and no audio came back. Why?

There are four usual causes. The bot was waiting for something else, such as a voice search or a key, and took your text as that: send /cancel and try again. The message began with a slash, so it was read as a command. Your previous audio was still being made. Or what you sent was not text, for example a photo with the words in it.

Where can I find audio I made in the bot later?

The History page on the website lists everything you made on the website and in the bot, and keeps the audio playable for 30 days. In the bot, /history lists your last ten conversions without the audio. Save a copy of anything you need for longer.

Ready to make your own?

Open the free converter, paste your script and download the MP3. No sign-up for short clips.

Open the converter

More guides

All guides →