Using Speech Search to speed editing

Transcribe spoken words to text and use Speech Search to find precise locations in clips. Speech Search makes editing clips of speaking talent quicker.

• View full-size on Adobe TV
• View more videos like this

Requirements

To complete the tasks demonstrated in this tutorial, you need the following software and files:

Adobe Premiere Pro CS4

• Try
• Buy

Prerequisite knowledge

Intermediate knowledge of sound editing in Adobe Premiere Pro

Using Speech Search to speed editing

Editing clips based on what is spoken can be a time-consuming process, frequently requiring you to play individual clips from beginning to end. To facilitate the editing process, Adobe® Premiere Pro® CS4 includes a Speech Search feature that allows you to search for what is actually spoken and to jump to a specific point in a recording. In this tutorial, you will learn how to create a transcript using the Speech Search feature and conduct a sound search across one or more files.

Creating a transcript using the Speech Search feature

The Speech Search feature in Adobe Premiere Pro CS4 makes it possible to create a searchable transcript of recorded speech. To use the Speech Search feature:

1. Open the sample file named lrvid4070_pr.prproj. Select Interview BTS 05.avi in the project panel. Make sure you have chosen the Effects workspace.
2. Choose Window > Metadata to open the Metadata panel.
3. To create a written transcription of the recording, click Transcribe in the lower-right corner of the Metadata panel. The Speech Transcription Options dialog box opens.

   

   Figure 1: The Metadata panel
4. In the Speech Transcription Options dialog box, leave the Language and Quality menus on their default settings. Leave Identify Speakers selected. Click OK. Adobe Media Encoder starts.
5. From the Adobe Media Encoder, click Start Queue to encode the project.

   Note: The Adobe Media Encoder encodes the file, takes everything that is spoken and transcribes it, and places it in the file as metadata. A green checkmark indicates the successful completion of the encoding process.

   

   Figure 2: The Adobe Media Encoder
6. Once the encoding process is complete, minimize the Adobe Media Encoder. Notice that the Speech Transcript portion of the Metadata panel now contains a transcript of the recording.
7. Play back the recording by clicking the Play-Stop toggle in the Source panel.

   

   Figure 3: The Play-Stop toggle
8. As the recording plays, Adobe Premiere Pro highlights in the Metadata panel each word as it is spoken. To jump to a specific word in the recording, highlight the word in the speech Transcript.
9. To edit a word in the transcript, click on the word. The word turns into a text field and allows you to type in corrections. Note that the editing process does not affect the timecode.

   

   Figure 4: Editing the speech Transcript

   Note: As you correct text, you are entering metadata that is encoded with the file. This process is especially helpful for sound editing. For example, you can use the Speech Transcript to set an inpoint and an outpoint in an interview.

Searching for words across multiple files

You can also use the Speech Search feature to locate words spoken across multiple files. To search for keywords:

1. Start Adobe Bridge.
2. Choose Edit > Find. The Find dialog box opens.
3. Under Criteria, make sure All Metadata is selected in the first text box and Contains is selected in the second text box. Type in the word for which you are searching in the third text box.

   

   Figure 5: Performing a word search
4. Click Find. Adobe Bridge conducts a search of every file that contains in its metadata the word you entered in the Search text box.

   Note: Edits made in Adobe Premiere Pro travel with that file, making it easy to locate files containing speech.

Where to go from here

• Adobe Premiere CS4 Help and Support
• Learn Adobe Premiere Pro CS4
• Introducing Adobe Premiere Pro CS4, by Paul Trani
• Viewing and editing metadata, by Paul Trani
• Normalizing audio for the whole Master track, by Paul Trani
• Lynda.com

This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License

About the authors