Query parameters
model=nova-3&language=en
Language Explorer
Nova-3 language controls
Idle
Language Explorer
Nova-3 Language Workbench
Search, explore, and apply Nova-3 language configurations across the lab.
Selected language
English
Code: en
- Model
- nova-3
- Configuration mode
- Explicit single-language transcription
- Base language
- English
- Regional variant
- Base language code
- Supported lab workflows
- Transcribe URL, Upload Audio, Live Mic, API Studio
- Verified metadata
- 2026-07-14
Recommended lab configuration
- audio known to be primarily English
- support calls conducted mainly in English
- English voice notes and recordings
Consider multilingual mode
- when speakers switch among languages in the verified multilingual list
- when the input language is unknown but expected within that list
- when a recording contains substantial code-switching
Known caveats from current project metadata
- Configuration guidance is not a benchmark result; validate representative audio before choosing a production default.
Configuration
Exact, copyable syntax for the project's verified prerecorded `/v1/listen` request style.
JSON configuration
{
"model": "nova-3",
"language": "en"
}TypeScript example
const audioBytes = await file.arrayBuffer();
const response = await fetch("https://api.deepgram.com/v1/listen?model=nova-3&language=en", {
method: "POST",
headers: {
Authorization: "Token YOUR_DEEPGRAM_API_KEY",
"Content-Type": file.type || "application/octet-stream",
},
body: audioBytes,
});
const transcript = await response.json();Python example
import requests
with open("audio.wav", "rb") as audio_file:
response = requests.post(
"https://api.deepgram.com/v1/listen?model=nova-3&language=en",
headers={
"Authorization": "Token YOUR_DEEPGRAM_API_KEY",
"Content-Type": "audio/wav",
},
data=audio_file,
timeout=60,
)
response.raise_for_status()
transcript = response.json()cURL binary-audio example
curl --request POST \ --url "https://api.deepgram.com/v1/listen?model=nova-3&language=en" \ --header "Authorization: Token YOUR_DEEPGRAM_API_KEY" \ --header "Content-Type: audio/wav" \ --data-binary @audio.wav
Apply this configuration
Navigation and prepopulation are local. Execution remains a separate visible action in the destination module.
English vs multilingual
Configuration guidance, not a benchmark result.
Explicit English
- • Use when English is known in advance.
- • More constrained configuration.
- • Best lab fit: primarily English audio.
Multilingual
- • Use when the input may switch among verified multilingual languages.
- • Broader recognition scope than one explicit code.
- • Best lab fit: mixed-language and unknown-language comparisons.
Applied Engineering Note
A starting hypothesis to validate with representative recordings.
Customer scenario
A support organization receives recordings primarily in English, with occasional product names or mixed-language phrases.
Recommended starting point
model: nova-3language: enAlternative
Use multilingual mode when substantial code-switching among verified languages is expected.
Validation plan
- test representative recordings
- compare only verified alternatives
- inspect names and domain terms
- verify regional audio where applicable
- evaluate latency and error patterns
- avoid choosing from one ideal sample
Sample text
Curated neutral fixture text; native-speaker or translation-quality certification is not claimed.
Hello, this is a short voice test for the Deepgram language lab.
Plain-English meaning: Hello, this is a short voice test for the Deepgram language lab.
Compare regional variants
These are distinct verified configuration values. This view does not claim how a model handles dialect or accent differences.
English
en
English (United States)
en-US
English (Australia)
en-AU
English (United Kingdom)
en-GB
English (India)
en-IN
English (New Zealand)
en-NZ