Transcribe audio files with the ML.TRANSCRIBE function

This document describes how to use the ML.TRANSCRIBE function with a remote model to transcribe audio files from an object table.

Supported locations

You must create the remote model used in this procedure in one of the following locations:

• asia-northeast1
• asia-south1
• asia-southeast1
• australia-southeast1
• eu
• europe-west1
• europe-west2
• europe-west3
• europe-west4
• northamerica-northeast1
• us
• us-central1
• us-east1
• us-east4
• us-west1

You must run the ML.TRANSCRIBE function in the same region as the remote model.

Required permissions

• To work with a Speech-to-Text recognizer, you need the following roles:

  • speech.recognizers.create
  • speech.recognizers.get
  • speech.recognizers.recognize
  • speech.recognizers.update

• To create a connection, you need membership in the following role:

  • roles/bigquery.connectionAdmin

• To create the model using BigQuery ML, you need the following permissions:

  • bigquery.jobs.create
  • bigquery.models.create
  • bigquery.models.getData
  • bigquery.models.updateData
  • bigquery.models.updateMetadata

• To run inference, you need the following permissions:

  • bigquery.tables.getData on the object table
  • bigquery.models.getData on the model
  • bigquery.jobs.create

Before you begin

1. Sign in to your Google Cloud account. If you're new to Google Cloud, create an account to evaluate how our products perform in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.

2. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

   Go to project selector

3. Make sure that billing is enabled for your Google Cloud project.

4. Enable the BigQuery, BigQuery Connection API, and Speech-to-Text APIs.

   Enable the APIs

5. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

   Go to project selector

6. Make sure that billing is enabled for your Google Cloud project.

7. Enable the BigQuery, BigQuery Connection API, and Speech-to-Text APIs.

   Enable the APIs

Create a recognizer

Speech-to-Text supports resources called recognizers. Recognizers represent stored and reusable recognition configurations. You can create a recognizer to logically group together transcriptions or traffic for your application.

Creating a speech recognizer is optional. If you choose to create a speech recognizer, note the project ID, location, and recognizer ID of the recognizer for use in the CREATE MODEL statement, as described in SPEECH_RECOGNIZER. If you choose not to create a speech recognizer, you must specify a value for the recognition_config argument of the ML.TRANSCRIBE function.

You can only use the chirp transcription model in the speech recognizer or recognition_config value that you provide.

Create a connection

Create a cloud resource connection and get the connection's service account.

Select one of the following options:

Console

1. Go to the BigQuery page.

   Go to BigQuery

2. To create a connection, click Add, and then click Connections to external data sources.

3. In the Connection type list, select Vertex AI remote models, remote functions and BigLake (Cloud Resource).

4. In the Connection ID field, enter a name for your connection.

5. Click Create connection.

6. Click Go to connection.

7. In the Connection info pane, copy the service account ID for use in a later step.

bq

1. In a command-line environment, create a connection:

   bq mk --connection --location=REGION --project_id=PROJECT_ID \
       --connection_type=CLOUD_RESOURCE CONNECTION_ID

   The --project_id parameter overrides the default project.

   Replace the following:

   • REGION: your connection region
   • PROJECT_ID: your Google Cloud project ID
   • CONNECTION_ID: an ID for your connection

   When you create a connection resource, BigQuery creates a unique system service account and associates it with the connection.

   Troubleshooting: If you get the following connection error, update the Google Cloud SDK:

   Flags parsing error: flag --connection_type=CLOUD_RESOURCE: value should be one of...
   

2. Retrieve and copy the service account ID for use in a later step:

   bq show --connection PROJECT_ID.REGION.CONNECTION_ID

   The output is similar to the following:

   name                          properties
   1234.REGION.CONNECTION_ID     {"serviceAccountId": "[e-mail]"}
   

Terraform

Append the following section into your main.tf file.

 ## This creates a cloud resource connection.
 ## Note: The cloud resource nested object has only one output only field - serviceAccountId.
 resource "google_bigquery_connection" "connection" {
    connection_id = "CONNECTION_ID"
    project = "PROJECT_ID"
    location = "REGION"
    cloud_resource {}
}        

Replace the following:

• CONNECTION_ID: an ID for your connection
• PROJECT_ID: your Google Cloud project ID
• REGION: your connection region

Grant access to the service account

Select one of the following options:

Console

1. Go to the IAM & Admin page.

   Go to IAM & Admin

2. Click Grant Access.

   The Add principals dialog opens.

3. In the New principals field, enter the service account ID that you copied earlier.

4. Click the Select a role field and then type Cloud Speech Client in Filter.

5. Click Add another role.

6. In the Select a role field, select Cloud Storage, and then select Storage Object Viewer.

7. Click Save.

gcloud

Use the gcloud projects add-iam-policy-binding command:

gcloud projects add-iam-policy-binding 'PROJECT_NUMBER' --member='serviceAccount:MEMBER' --role='roles/speech.client' --condition=None
gcloud projects add-iam-policy-binding 'PROJECT_NUMBER' --member='serviceAccount:MEMBER' --role='roles/storage.objectViewer' --condition=None

Replace the following:

• PROJECT_NUMBER: your project number.
• MEMBER: the service account ID that you copied earlier.

Failure to grant the permission results in a Permission denied error.

Note: If you create the recognizer in a different project than the Cloud Storage bucket used by the object table, grant the service account Identity and Access Management (IAM) roles as follows:

• Grant the service account the Cloud Speech Client role in the project that contains the recognizer.
• Grant the service account the Storage Object Viewer role in the project that contains the Cloud Storage bucket.
• Grant the Speech-to-Text service agent ([e-mail]) the Storage Object Viewer role in the project that contains the Cloud Storage bucket.

Create a dataset

Create a dataset to contain the model and the object table.

Create an object table

Create an object table over a set of audio files in Cloud Storage. The audio files in the object table must be of a supported type.

The Cloud Storage bucket used by the object table should be in the same project where you plan to create the model and call the ML.TRANSCRIBE function. If you want to call the ML.TRANSCRIBE function in a different project than the one that contains the Cloud Storage bucket used by the object table, you must grant the Storage Admin role at the bucket level to the [e-mail] service account.

Create a model

Create a remote model with a REMOTE_SERVICE_TYPE of CLOUD_AI_SPEECH_TO_TEXT_V2:

CREATE OR REPLACE MODEL
`PROJECT_ID.DATASET_ID.MODEL_NAME`
REMOTE WITH CONNECTION `PROJECT_ID.REGION.CONNECTION_ID`
OPTIONS (
  REMOTE_SERVICE_TYPE = 'CLOUD_AI_SPEECH_TO_TEXT_V2',
  SPEECH_RECOGNIZER = 'projects/PROJECT_NUMBER/locations/LOCATION/recognizers/RECOGNIZER_ID'
);

Replace the following:

• PROJECT_ID: your project ID.
• DATASET_ID: the ID of the dataset to contain the model.
• MODEL_NAME: the name of the model.
• REGION: the region used by the connection.
• CONNECTION_ID: the connection ID—for example, myconnection.

  When you view the connection details in the Google Cloud console, the connection ID is the value in the last section of the fully qualified connection ID that is shown in Connection ID—for example projects/myproject/locations/connection_location/connections/myconnection.

• PROJECT_NUMBER: the project number of the project that contains the speech recognizer. You can find this value on the Project info card in the Dashboard page of the Google Cloud console.
• LOCATION: the location used by the speech recognizer. You can find this value in the Location field on the List recognizers page of the Google Cloud console.
• RECOGNIZER_ID: the speech recognizer ID. You can find this value in the ID field on the List recognizers page of the Google Cloud console.

  This option isn't required. If you don't specify a value for it, a default recognizer is used. In that case, you must specify a value for the recognition_config parameter of the ML.TRANSCRIBE function in order to provide a configuration for the default recognizer.

  You can only use the chirp transcription model in the recognition_config value that you provide.

Important: You must specify the project ID for the connection even if the connection is in the default project.

Transcribe audio files

Transcribe audio files with the ML.TRANSCRIBE function:

SELECT *
FROM ML.TRANSCRIBE(
  MODEL `PROJECT_ID.DATASET_ID.MODEL_NAME`,
  TABLE `PROJECT_ID.DATASET_ID.OBJECT_TABLE_NAME`,
  RECOGNITION_CONFIG => ( JSON 'recognition_config')
);

Replace the following:

• PROJECT_ID: your project ID.
• DATASET_ID: the ID of the dataset that contains the model.
• MODEL_NAME: the name of the model.
• OBJECT_TABLE_NAME: the name of the object table that contains the URIs of the audio files to process.
• recognition_config: a RecognitionConfig resource in JSON format.

  If a recognizer has been specified for the remote model by using the SPEECH_RECOGNIZER option, you can't specify a recognition_config value.

  If no recognizer has been specified for the remote model by using the SPEECH_RECOGNIZER option, you must specify a recognition_config value. This value is used to provide a configuration for the default recognizer.

  You can only use the chirp transcription model in the recognition_config value that you provide.

Examples

Example 1

The following example transcribes the audio files represented by the audio table without overriding the recognizer's default configuration:

SELECT *
FROM ML.TRANSCRIBE(
  MODEL `myproject.mydataset.transcribe_model`,
  TABLE `myproject.mydataset.audio`
);

The following example transcribes the audio files represented by the audio table and provides a configuration for the default recognizer:

SELECT *
FROM ML.TRANSCRIBE(
  MODEL `myproject.mydataset.transcribe_model`,
  TABLE `myproject.mydataset.audio`,
  recognition_config => ( JSON '{"language_codes": ["en-US" ],"model": "chirp","auto_decoding_config": {}}')
);

What's next

• For information about model inference in BigQuery ML, see Model inference overview.
• For information about the supported SQL statements and functions for each model type, see End-to-end user journey for each model.