Deploying substitutions

Substitution files must be deployed before they can be used as part of an automated speech-to-text processing flow. The substitution files are deployed by specifying a substitution file or substitution rules to be applied during transcription. The process varies depending on the ASR solution you are using. The following sections explain the deployment process for each Voci ASR solution.

Deploying substitutions to V‑Blaze

V‑Blaze is an ASR engine that performs speech-to-text transcription and provides user access via a RESTful API. The following steps describe how to deploy substitutions on V‑Blaze .

Place the substitution file in a working directory.
Use the following cURL command to submit audio for transcription with substitutions applied.

mySubFile.sub in the following command should be set to the name of your substitution file.

curl -F 'subst_rules=</path/mySubFile.sub' -F 'model=eng1:callcenter' -F 'output=text' -F 'file=@example/sample1.wav' http://blaze_name:17171/transcribe

The following list describes the parameter specifications from the cURL command above.

subst_rules: Specifies a newline-delimited string that contains substitution rules to be applied during transcription. The substitution rules are specified in a local file and loaded into the subst_rules field by using the < operator as shown in the curl command above. This approach makes it easier to edit and maintain these rules over time.
model: Specifies the language model to be used during transcription. Use the same language model previously used during the development of the substitution rules.
output: The transcription output format. Text is specified here for readability.
file: Audio file to be transcribed.

Deploying substitution rules to V‑Cloud

V‑Cloud is Voci 's cloud-based Speech-to-Text transcription service.

Use the following commands to submit a transcription request on V‑Cloud. Each command should be on a single line.

Set a variable with a valid token

token='token-here'

Set a variable to specify an audio file to transcribe

filename='/path/sample1.wav'

mySubFile.sub in the following command should be set to your substitution file.

Submit audio for transcription and capture the returned request ID

requestID=$(curl -F 'subst_rules=</path/mySubFile.sub' -F 'model=eng1:callcenter' -F 'output=text' -F "token=$token" -F "file=@$filename" https://vcloud.name.com/transcribe | cut -d\" -f4)

Command to get download link for results

response=$(curl "https://vcloud.vocitec.com/transcribe/result?token=$token&requestid=$requestID")

Check if the "response" variable contains a URL using the following command.

echo $response

Repeat the previous cURL command until "response" contains a URL.

Command to download results

curl -o result.txt "$(cut -d\" -f2 <<< $response)"

Check the results to make sure that substitutions were successfully applied.