Docs / Deepgram-compatible API / Plugins

Deepgram-compatible API

Plugins

Pipecat has a .wave speech-to-text service, pipecat-dotwave. LiveKit Agents, and Pipecat pipelines already built on Deepgram, connect through their Deepgram plugin by changing its base URL. Either way, the rest of the pipeline stays as it is.

LiveKit Agents

import os
from livekit.plugins import deepgram

stt = deepgram.STT(
    api_key=os.environ["DOTWAVE_API_KEY"],
    base_url="https://api.dotwave.ai/v1/listen",
    model="nemotron-asr-streaming",
    language="pt-BR",
)

Pipecat

Install pipecat-dotwave and put the service after your voice detection. It pushes one transcription per segment, with interim results carrying the segment so far, and sends Finalize when the detector reports the end of speech. The package page has the full example and the options.

uv add pipecat-dotwave
import os
from pipecat.transcriptions.language import Language
from pipecat_dotwave import WaveSTTService

stt = WaveSTTService(
    api_key=os.environ["DOTWAVE_API_KEY"],
    settings=WaveSTTService.Settings(language=Language.PT_BR),  # leave it out to detect the language
)

A pipeline already built on Pipecat’s Deepgram service can keep it and change two settings instead:

import os
from pipecat.services.deepgram.stt import DeepgramSTTService
from pipecat.transcriptions.language import Language

stt = DeepgramSTTService(
    api_key=os.environ["DOTWAVE_API_KEY"],
    base_url="api.dotwave.ai",
    settings=DeepgramSTTService.Settings(
        model="nemotron-asr-streaming",
        language=Language.PT_BR,
    ),
)

Settings

Model
Can stay at the plugin’s default.
Language
Name it when you know it; leave it out to let the model detect it. See Languages.
Other settings
Plugin defaults such as punctuate, smart_format and interim_results are accepted. See Parameters.