Pronunciation Dictionary

If your agents work with names, places, or terms your text-to-speech provider doesn't pronounce correctly by default, you can add a pronunciation dictionary entry to fix it — no per-workflow setup required. One shared dictionary is used by every workflow in your organization.

How it works

1. You add entries — a term (the exact text as it appears in your agent's replies) and a sounds-like respelling, with an optional advanced IPA field for finer control.
2. Talkr automatically pushes your dictionary to your configured TTS provider in the background whenever it changes. There's no separate publish step per workflow.
3. Every agent call for your organization picks up the current dictionary automatically.

Adding an entry

1. Go to Settings → Pronunciation Dictionary.
2. Enter the term exactly as it will appear in your agent's replies — matching is case-sensitive.
3. Enter a sounds-like respelling — plain phonetic spelling, e.g. ben-ga-LOO-ru for "Bengaluru".
4. Optionally expand Advanced and enter raw IPA for more precise phonetic control.
5. Click add.

Editing an entry

Click the pencil next to an entry to change its term, sounds-like respelling, or IPA in place, then Save. Clearing the IPA field removes the IPA override and falls back to the sounds-like respelling.

Editing re-syncs the dictionary to your TTS provider once. Previously the only way to correct an entry was to delete and re-add it, which pushed two syncs.

Removing an entry

Click the bin next to an entry. The dictionary re-syncs to your provider automatically.

Checking sync status

A badge next to each provider shows whether your dictionary is up to date:

• Synced — the dictionary is current on that provider.
• Syncing — an update is in progress; this usually resolves within moments.
• Sync error — the last push failed (for example, an invalid API key). Your previous working dictionary keeps applying to calls until the error is fixed; use Retry sync once you've corrected the issue.
• Not in use — this provider isn't your organization's currently configured TTS provider, so its dictionary isn't kept up to date.

Related

• Voice — configure your TTS provider
• Voice & Language Profiles — per-language voice switching