Smart Formatting

Smart Format formats transcripts to improve readability.

Deepgram's Smart Format feature applies additional formatting to transcripts to optimize them for human readability.

Smart Format capabilities vary between models. When Smart Format is turned on, Deepgram will always apply the best-available formatting for your chosen combination of model, tier, and language.

At minimum, Smart Format applies:

Smart Format has the broadest support for English-language models.

  • When using the English Nova or Enhanced general models, Smart Format is capable of formatting things like dates, times, currency amounts, phone numbers, emails, and URLs.
  • On other English models, Smart Format will format a smaller set of entities, including dates, times, and numbers.

On non-English models, Smart Format will apply all available formatters for that language. For the majority of languages, this is currently limited to Punctuation and Paragraphs.

Enable Feature

To enable Smart Formatting, when you call Deepgram's API, add a smart_format parameter in the querystring and set it to true:

smart_format=true

๐Ÿ“˜

Smart Format enables Deepgram's Punctuation feature. If you've set smart_format=true, no need to also set punctuate=true.

To transcribe audio from a file on your computer, run the following cURL command in a terminal or your favorite API client.

curl \
  --request POST \
  --header 'Authorization: Token YOUR_DEEPGRAM_API_KEY' \
  --header 'Content-Type: audio/wav' \
  --data-binary @youraudio.wav \
  --url 'https://api.deepgram.com/v1/listen?smart_format=true'

:eyes: Replace YOUR_DEEPGRAM_API_KEY with your Deepgram API Key.

Results

Once applied, results will appear in the transcript.

TruthBefore Smart FormatAfter Smart Format
i'm recording this at eight thirty seven pm on wednesday it's november second twenty twenty two i just ate thirty three grams of pasta and then drank fifty five milliliters of water at three hundred main street then i walked down to one two three southeast main street to get my package with tracking number one z five seven a two bi'm recording this at eight thirty seven pm on wednesday it's november second twenty twenty two i just ate thirty three grams of pasta and then drank fifty five milliliters of water at three hundred main street then i walked down to one two three southeast main street to get my package with tracking number one z five seven a two bI'm recording this at 8:37 PM on Wednesday, it's November 2 2022. I just ate 33 grams of pasta, and then drank 55 milliliters of water at 300 Main Street.

Then I walked down to 123 Southeast Main Street to get my package with tracking number 1Z57A2B.

๐Ÿ“˜

When using Smart Format with live-streamed audio, if a speaker begins saying a number, Deepgram will wait to return a transcription until the speaker has finished and continues on to non-numerical speech. This behavior ensures numbers have the best possible formatting and are not broken up over multiple chunks. To override this behavior and return results immediately, add the parameter no_delay=true to your streaming API request.

Use Cases

Examples of use cases for Smart Format include:

  • Customers who want to improve transcript readability.
  • Customers who want to search transcripts for specific types of entities to perform data analysis.
  • Customers who want to build custom redaction functional that only operates on specific types of entities.

Additional Formatters

These formatters are not included in Smart Formatting but may be enabled individually.

Measurements

If Measurements is enabled, spoken measurements will be converted to their corresponding abbreviations.

To enable Measurements, when you call Deepgram's API, add measurements=true to your request.

The following units will be converted to their abbreviations.

UnitAbbreviation
milligram(s)mg
centigram(s)cg
gram(s)g
kilogram(s)kg
milliliter(s)ml
centiliter(s)cl
liter(s)l
kiloliter(s)kl
millimeter(s)mm
centimeter(s)cm
meter(s)m
kilometer(s)km

Dictation

If Dictation is enabled, spoken dictation commands will be converted to their corresponding punctuation marks.

To enable Dictation, when you call Deepgram's API, add dictation=true&punctuate=true to your request. Punctuation must be enabled for Dictation to work.

The following commands will be converted.

CommandConverted Text
period.
comma,
colon:
question mark?
exclamation mark!
new paragraph<\n>

โ„น๏ธ

By default, Deepgram applies its Base tier, general AI model, which is a good, general-purpose model for everyday situations. To learn more about the customization possible with Deepgram's API, check out the Deepgram API Reference.