Bhashini

If your business operates in Indian languages, most transcripts you already have are useful to exactly one person: whoever was on the call. Bhashini makes them useful to a wider audience — a reviewer who only reads English, a manager who works in a different regional language than your callers. The same credential also lets your CRM show contact names in their native script instead of however they were typed into a spreadsheet.

What it is

Bhashini is the Government of India's National Language Translation Mission — a free, production API covering translation, transcription and related language services across 22 Indian languages. Talkr connects to it using your organization's own Bhashini credentials, so usage is attributed to your account, not shared across every Talkr customer.

Once connected, the same credential unlocks two things:

• Transcript translation — a completed call's transcript is translated into a language you choose, and attached to the call record (see "Setting it up" below)
• CRM name transliteration — when importing contacts, you can convert names from Latin script into a regional script, so "vishal gupta" becomes "विशाल गुप्ता" on the board. This is opt-in per import, ticked on the import screen itself — see the CRM guide for how.

Getting your credentials

Bhashini credentials come from the Bhashini dashboard, not from Talkr:

1. Register at bhashini.gov.in and sign in
2. From My Profile, generate an API key for your application
3. Note your User ID and Udyat key (also called the ULCA API key) — both shown on that same page

Neither of these is a secret you share with anyone else — they identify your organization to the Government of India's platform, and Talkr stores them encrypted, the same way it stores every other credential you connect.

Setting it up

Step 1 is the same for both translation and CRM transliteration: in Settings → Bhashini, paste your User ID and Udyat key and save. Talkr validates them against Bhashini's own servers before saving — if the key is wrong, you'll be told immediately rather than discovering it later when a call or an import fails.

CRM transliteration is ready to use as soon as credentials are saved — turn it on from the import screen itself when you next upload a contact list (see the CRM guide).

Transcript translation needs two more steps, back in Settings → Bhashini:

1. Set your source language — the ISO language code your calls are actually spoken in (hi for Hindi, mr for Marathi, ta for Tamil, and so on) — and the language you want transcripts translated into.
2. Turn translation on and save.

Where the results show up

The translated transcript is attached to the call record under its own section, alongside the transcript, recording and QA results you already see there. A call that hasn't finished translating yet — or one where translation failed — simply shows nothing extra; the rest of the call record is unaffected either way.

Translation runs as a background step after the call ends, typically within moments of the call completing. It never delays anything the caller experiences, and it can't fail a call — a Bhashini outage means that call's transcript stays un-translated, not that the call itself is affected in any way.

What this doesn't do

It doesn't change how your agent sounds or responds on a live call. Bhashini translation is a batch, after-the-fact process — it reads a finished transcript and writes back a translated copy. Your agent's speech recognition and voice continue to run on whichever providers you've configured for the live call (Settings → Model Configuration), completely independently of this.

If you're looking to change what language or voice your agent uses during a call, that's a different setting — see Model Configurations and Voice & Language Profiles.