Edit

Configure the Speech CLI output options

  • Article

The Speech CLI output can be written to standard output or specified files.

For contextual help in the Speech CLI, you can run any of the following commands:

spx help recognize output examples
spx help synthesize output examples
spx help translate output examples
spx help intent output examples

Standard output

If the file argument is a hyphen (-), the results are written to standard output as shown in the following example.

spx recognize --file caption.this.mp4 --format any --output vtt file - --output srt file - --output each file - @output.each.detailed --property SpeechServiceResponse_StablePartialResultThreshold=0 --profanity masked

Default file output

If you omit the file option, output is written to default files in the current directory.

For example, run the following command to write WebVTT and SRT captions to their own default files:

spx recognize --file caption.this.mp4 --format any --output vtt --output srt --output each text --output all duration

The default file names are as follows, where the <EPOCH_TIME> is replaced at run time.

  • The default SRT file name includes the input file name and the local operating system epoch time: output.caption.this.<EPOCH_TIME>.srt
  • The default Web VTT file name includes the input file name and the local operating system epoch time: output.caption.this.<EPOCH_TIME>.vtt
  • The default output each file name, each.<EPOCH_TIME>.tsv, includes the local operating system epoch time. This file isn't created by default, unless you specify the --output each option.
  • The default output all file name, output.<EPOCH_TIME>.tsv, includes the local operating system epoch time. This file is created by default.

Output to specific files

For output to files that you specify instead of the default files, set the file option to the file name.

For example, to output both WebVTT and SRT captions to files that you specify, run the following command:

spx recognize --file caption.this.mp4 --format any --output vtt file caption.vtt --output srt file caption.srt --output each text --output each file each.result.tsv --output all file output.result.tsv

The preceding command also outputs the each and all results to the specified files.

Output to multiple files

For translations with spx translate, separate files are created for the source language (such as --source en-US) and each target language (such as --target "de;fr;zh-Hant").

For example, to output translated SRT and WebVTT captions, run the following command:

spx translate --source en-US --target "de;fr;zh-Hant" --file caption.this.mp4 --format any --output vtt file caption.vtt --output srt file caption.srt

Captions should then be written to the following files: caption.srt, caption.vtt, caption.de.srt, caption.de.vtt, caption.fr.srt, caption.fr.vtt, caption.zh-Hant.srt, and caption.zh-Hant.vtt.

Suppress header

You can suppress the header line in the output file by setting the has header false option:

spx recognize --nodefaults @my.defaults --file audio.wav --output recognized text --output file has header false

See Configure the Speech CLI datastore for more information about --nodefaults.

Next steps