DEV Community

Cover image for Introducing Whisper
Prathamesh Belvalkar
Prathamesh Belvalkar

Posted on

Introducing Whisper

Whisper is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitask supervised data collected from the web. We show that the use of such a large and diverse dataset leads to improved robustness to accents, background noise and technical language. Moreover, it enables transcription in multiple languages, as well as translation from those languages into English. We are open-sourcing models and inference code to serve as a foundation for building useful applications and for further research on robust speech processing.

Whisper Architecture

The Whisper architecture is a simple end-to-end approach, implemented as an encoder-decoder Transformer. Input audio is split into 30-second chunks, converted into a log-Mel spectrogram, and then passed into an encoder. A decoder is trained to predict the corresponding text caption, intermixed with special tokens that direct the single model to perform tasks such as language identification, phrase-level timestamps, multilingual speech transcription, and to-English speech translation.

How it works

Other existing approaches frequently use smaller, more closely paired audio-text training datasets,1 2, 3 or use broad but unsupervised audio pretraining.4, 5, 6 Because Whisper was trained on a large and diverse dataset and was not fine-tuned to any specific one, it does not beat models that specialize in LibriSpeech performance, a famously competitive benchmark in speech recognition. However, when we measure Whisper’s zero-shot performance across many diverse datasets we find it is much more robust and makes 50% fewer errors than those models.

About a third of Whisper’s audio dataset is non-English, and it is alternately given the task of transcribing in the original language or translating to English. We find this approach is particularly effective at learning speech to text translation and outperforms the supervised SOTA on CoVoST2 to English translation zero-shot.

ASR training data inputs and outputs

We hope Whisper’s high accuracy and ease of use will allow developers to add voice interfaces to a much wider set of applications. Check out the paper(opens in a new window), model card(opens in a new window), and code(opens in a new window) to learn more details and to try out Whisper.

Billboard image

Deploy and scale your apps on AWS and GCP with a world class developer experience

Coherence makes it easy to set up and maintain cloud infrastructure. Harness the extensibility, compliance and cost efficiency of the cloud.

Learn more

Top comments (0)

Cloudinary image

Optimize, customize, deliver, manage and analyze your images.

Remove background in all your web images at the same time, use outpainting to expand images with matching content, remove objects via open-set object detection and fill, recolor, crop, resize... Discover these and hundreds more ways to manage your web images and videos on a scale.

Learn more

👋 Kindness is contagious

Explore a sea of insights with this enlightening post, highly esteemed within the nurturing DEV Community. Coders of all stripes are invited to participate and contribute to our shared knowledge.

Expressing gratitude with a simple "thank you" can make a big impact. Leave your thanks in the comments!

On DEV, exchanging ideas smooths our way and strengthens our community bonds. Found this useful? A quick note of thanks to the author can mean a lot.

Okay