Navana

Navana is a speech-to-text plugin. It uses Navana’s Bodhi API and suits Indian languages and 8 kHz phone-call audio.

Setup

Set your Navana API key in the worker environment. Generate a key from the Navana site:

export NAVANA_API_KEY=<key>
export NAVANA_CUSTOMER_ID=<customer-id>

Import the plugin and pass it to the pipeline’s stt slot.

from zeroruntime.plugins import NavanaSTT

stt = NavanaSTT()

# Pipeline(stt=stt, ...)

import { NavanaSTT } from '@zeroruntime/js-sdk/plugins';

const stt = NavanaSTT();

// Pipeline({ stt, ... })

Constructor parameters for NavanaSTT.

Parameter	Type	Default	Description
`api_key`	`str`	`None`	Navana API key. Falls back to the `NAVANA_API_KEY` environment variable when unset.
`customer_id`	`str \| None`	`None`	Your Navana customer ID, issued with your API key.
`model`	`str`	`"en-general-v2-8khz"`	Bodhi model name, e.g. “en-general-v2-8khz”. Pick one that matches the language and audio sample rate.
`language`	`str`	`"en"`	Language code of the speech, e.g. “en” or “hi”.
`input_sample_rate`	`int \| None`	`None`	Sample rate of incoming audio in Hz. Leave unset to detect it automatically.
`interim_results`	`bool`	`True`	Also send partial transcripts to the client app. Partial results are always used for interruption and turn detection either way.

SDK	Import	Constructor
Python	`from zeroruntime.plugins import NavanaSTT`	`NavanaSTT()`
Node JS	`import { NavanaSTT } from '@zeroruntime/js-sdk/plugins'`	`NavanaSTT()`

Transcribed text passes to the LLM once turn detection decides the caller has finished speaking.