NVIDIA Canary: multilingual transcription and translation models

by NVIDIA

Don’t want to install NVIDIA Canary? Drop the file here.

The first 20 minutes of any recording are transcribed free: a recording up to 20 minutes transcribes in full, a longer one is previewed, and a pack from $6 (500 minutes) transcribes the rest and opens every export, and credits never expire.

Canary models transcribe and translate speech: canary-1b-flash covers four languages at over 1,000× real time, canary-1b-v2 covers 25 European languages with translation to and from English.

Linux  ·  free

Get NVIDIA Canary Official site & pricing

TL;DR

Canary models do transcription and speech translation: canary-1b-flash (4 languages, 1,000×+ real time) and canary-1b-v2 (25 European languages, to and from English). CC-BY-4.0.

Sibling of Parakeet in NeMo. Clips over 40 seconds need chunking; a hosted service takes whole files.

Category
Open source
License
CC-BY-4.0
Pricing
free (CC-BY-4.0)
Platforms
Linux

What it is

Canary is NVIDIA's family of FastConformer encoder and Transformer decoder models in NeMo. canary-1b-flash (883M parameters) transcribes English, German, Spanish and French, translates between English and those languages, and runs at more than 1,000× real time on the Open ASR Leaderboard data. canary-1b-v2 (978M parameters, August 2025) extends this to 25 European languages, with translation from English into 24 languages and from 24 into English, and word- and segment-level timestamps. Both are CC-BY-4.0 and the cards state they are ready for commercial use. Canary models are designed for audio under 40 seconds; NeMo's chunked inference script handles longer files.

Best for: Teams that need transcription and speech translation in European languages from one open, commercially usable model on NVIDIA GPUs.
Watch out for: Built for clips under 40 seconds: longer audio goes through NeMo's chunked-inference script. European languages only, and optimised for NVIDIA GPUs.

Whole files, no chunking

  1. Drop the audio or video file into the box at the top of this page, or paste a link (YouTube, a podcast or RSS feed, Google Drive, Twitch or Kick). Nothing to install and no GPU needed.
  2. The language is detected automatically and the transcript stays in the language that was spoken (WhipScribe does not translate).
  3. Read the summary free. A pack from $6 (500 minutes) opens the full transcript and the downloads: TXT, DOCX, SRT, VTT and JSON. Packs are one-time purchases and credits never expire.
  4. Files can be up to 10 hours or 5 GB per file.

Related: Bulk transcription · SRT generator · Transcribe large files · WhisperX · Pricing

Install / use

pip install -U "nemo_toolkit[asr]"  # model: nvidia/canary-1b-v2

Key specs

canary-1b-v2
978M · 25 languages
canary-1b-flash
883M · 4 languages
Tasks
Transcription + translation
Licence
CC-BY-4.0

What people actually do with NVIDIA Canary-style transcription

The tool is the means. These are the jobs — each one priced at published rates, each one wired up on its own page.

Transcribe a whole back catalogue in one runThe archive is the asset, but one-at-a-time upload means it never gets done.…1,000 hours ≈ $480, onceTranscribe every new file in a Drive folderFiles accumulate in a folder nobody opens, and the knowledge in them is invisible to search.…≈ $0.29 per audio hour on the $24 packCut short-form clips out of a webinar automaticallyThe clips worth posting are buried somewhere in a 60-minute recording and finding them by scrubbing costs more than the…60-min webinar ≈ $0.72

Wire it into Amazon S3n8nMake.comGoogle DriveZapierSlack

Features

Speaker diarizationNot stated
Word-level timestampsYes
Streaming / real-timeNot stated
Languages supported25
HIPAA eligibleNo

Links

NVIDIA Canary vs Whipscribe

FeatureNVIDIA CanaryWhipscribe
CategoryOpen sourceTranscription APIs
Pricingfree (CC-BY-4.0)$6–$24 one-time packs (credits never expire) · first 20 minutes free
Speaker identificationNoNamed in the summary, key quotes and chapters
Word timestampsYesYes
StreamingNoNo
Languages2599
PlatformsLinuxWeb, API, MCP

Alternatives to NVIDIA Canary

OpenAI Whisper
OpenAI

The reference open-source multilingual ASR model from OpenAI.

OSS · MIT ★ 108.7k
whisper.cpp
Georgi Gerganov

C/C++ port of Whisper — runs on anything, from a Raspberry Pi to Apple Silicon.

OSS · MIT ★ 53.5k
faster-whisper
SYSTRAN

4× faster than reference Whisper using CTranslate2 — production sweet spot.

OSS · MIT ★ 25.3k

Frequently asked about NVIDIA Canary

What is the difference between Canary and Parakeet?

Canary is an encoder-decoder model that can also translate speech; Parakeet is a transcription-only model built for speed. Both are NVIDIA NeMo models under CC-BY-4.0.

Which Canary model should I use?

canary-1b-v2 for the 25 European languages and translation in both directions; canary-1b-flash for the fastest transcription in English, German, Spanish and French.

Can Canary transcribe long audio?

The models are designed for input under 40 seconds. NeMo's speech_to_text_aed_chunked_infer.py script splits longer audio, transcribes the chunks and stitches the result.

Does WhipScribe translate like Canary does?

No. WhipScribe transcribes in the language that was spoken and does not translate. If you need English subtitles for German audio, Canary's translation task does that.

Is NVIDIA Canary free?

Yes. NVIDIA Canary is free and open source (free (CC-BY-4.0)).

Does NVIDIA Canary work on Mac, Windows and Linux?

NVIDIA Canary runs on Linux. It is not listed for other platforms; check the vendor for anything beyond Linux.

How do I install NVIDIA Canary?

With pip: `pip install -U "nemo_toolkit[asr]" # model: nvidia/canary-1b-v2`. You need Python and, for most audio tools, ffmpeg on your PATH first.

How many languages does NVIDIA Canary support?

NVIDIA Canary lists 25 languages.

Does NVIDIA Canary give word-level timestamps?

Yes — NVIDIA Canary produces timing per word, which is what subtitle cues and clip boundaries need.

Is NVIDIA Canary HIPAA compliant?

NVIDIA Canary does not claim HIPAA compliance. For PHI, run it on infrastructure you control or choose a service that will sign a BAA.

What are the limitations of NVIDIA Canary?

Built for clips under 40 seconds: longer audio goes through NeMo's chunked-inference script. European languages only, and optimised for NVIDIA GPUs.

Who is NVIDIA Canary best for?

Teams that need transcription and speech translation in European languages from one open, commercially usable model on NVIDIA GPUs.

What should I know before choosing NVIDIA Canary?

Sibling of Parakeet in NeMo. Clips over 40 seconds need chunking; a hosted service takes whole files.

Does NVIDIA Canary run offline?

NVIDIA Canary runs on your own machine — it is open source, so nothing leaves the computer unless you configure it to.

Can I use NVIDIA Canary in the browser?

No — NVIDIA Canary is a Linux application, not a web app. You install it rather than sign in to it.

Whipscribe is a hosted transcription service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.

Transcribe your recordings

Running NVIDIA Canary yourself? Send the same file here and compare the transcript. Transcribe any language, any recording · Transcription services from $0.29 an hour · Transcribe from any meeting or recording platform · How to transcribe, step by step

Explore

All transcription tools · By language · By platform · By file format · How-to guides · Transcription services · Transcription API · Pricing

Have the recording already?

Skip the setup and send the file or the link: see how to transcribe an MP3 of any length, a podcast episode as text from its link and the SRT generator.

Comparing tools before you pay?

If you would rather pay a service than run software yourself: see an honest roundup of the best transcription services, affordable transcription priced per hour and how our transcription services work.