MacSpeech Scribe is speech recognition software for Mac OS X designed specifically for transcription of recorded voice dictation.
In order to get started you need to already have a 120 second long audio recording (.wav, .aif, .aiff, .mp4, .m4a, .m4v) of someone reading text. This profile will belong to them. The person must be reading text so that you can reference the material during the calibration portion.
Create a Profile
1) Open MacSpeech Scribe.
2) Press the + button to create a new profile. Save your profile in a place you can find it again. - fig.1
3) After saving your profile you will be prompted with a few choices - choose US for spelling and American accent from the list. - fig.2
Note: After you click Create your decisions are permanent and you would need to create a new profile in order to make changes.
4) You will be prompted to find your 120 second long recording.
5) Again you will be prompted with a window that describes how the process of Transcription Training works. You will press transcribe and follow along with the software. - fig.3
6) Scribe will begin attempting to transcribe your audio. It will break up the recording in to sections based on pauses detected in your voice. These will be highlighted in blue when you scroll over them. - fig.4
7) Click your first line, and read through the transcription one line at a time. Accept the successful transcriptions. For the unsuccessful transcriptions, make edits before you submit them. - fig.5
8) After you have gone through your text and edited the mistakes, click Calibrate. The process will begin again- this time with less mistakes.
9) Continue the transcription training until you have finished the 120 second audio clip. You can keep training if your clip is longer, this will make your profile and MacSpeech Scribe's transcriptions more accurate.
From the main window you can select Vocabulary Training... From this MacSpeech Scribe can learn how you string words together which will enhance it's ability to understand you. The more similar the documents are (content wise) to the audio, the more useful they will be.
1) Open Vocabulary Training...
2) Click the Add button at the bottom left of the screen and add as many word documents you have written as you want. - fig.6
3) Press the right arrow button after you have finished selecting.
4) You will then be prompted to choose words to include in the vocabulary training and words to exclude. - fig.7
5) Press the right arrow button after you have finished your selection.
6) Even if there are no new words detected, you should still press the right arrow to continue. The order of the words within the document still aid in improving the transcriptions.
7) You have finished your vocabulary training. A window will tell you can come back and do more training through the Tools menu or from the main window.
- Vocabulary Editor is a list of words that MacSpeech Scribe understands and is able to transcribe for you. When you submit word documents it pulls vocabulary it doesn't know from there. When you make edits during the calibration process these changes are added to the vocabulary list. - fig.8
- You can view the complete list All, the list included with the software Built-In, or the list created during the process of building your profile User. For each word in the list you can attach information and aid in enhancing MacSpeech Scribe's ability to use it.
Note: If you are consistently having to correct words with in your text for capitalization or spelling you should visit this portion of the Vocabulary Editor. Here you can customize each word in a variety of ways. - fig.9
- You can also add vocabulary words manually. You add a word by selecting the + button on the bottom left of the window (see above). A window will appear, and you add words as you see necessary. You can delete items from the Vocabulary Editor by right clicking the word. - fig.10
Importing/Exporting a Vocabulary List
1) This may be useful for adding vocabulary from one profile to another, and may increase your efficiency at building accurate profiles. To do this you will need to select the words you intend to export. and under the Gear menu press export. - fig.11
2) You will then be prompted identify a location to save the file. The file will export as a XML file. In you can import the vocabulary list in to a new profile through the Vocabulary Editor and the Gear menu.
1) You can begin the transcription process after you have completed the suggested 120 second training.
Note: it is encouraged you do more than 120 seconds of training to build a stronger profile. Also it is a good idea to upload several documents to the vocabulary training section. This is especially encouraged if the documents contain technical vocabulary that could be used within the audio being transcribed. For example, if you are attempting to record and transcribe the process of building a color environment through ColorBurst, it would be a good idea to submit documents use similar vocabulary to that of the audio. Scribe will be able to learn the technical vocabulary and implement its knowledge during the transcription process.
2) You can access the Transcription window either from the main window, or directly from the Transcription Training... window. - fig.12
3) After you begin the transcription process, a new window will open and within a matter of moments the transcription process will begin. This process will not be perfect and you will need to continue editing. However, in this window there will be suggestions on how to correct the transcription. These suggestions are found in the Recognition portion of the window. Click the number for the alternative you would like to add. - fig.13
4) If none of the suggestions work then you edit the entry by double clicking it. Your edit will show up in the Recognition portion of the window. Double click it to insert it in to the transcription.
Keyboard Shortcuts for the Transcription Process
Exporting the Transcribed Text
1) Simply copy and paste the text from the transcription window in to a word document, or any application that can accept text.
2) You can also save the transcription through the file menu using Save or Save As. The file will be saved as a Rich Text Format (.rtf), which can then be imported in to many word processing applications. These applications handle the formatting for the information.
1) In situations where you are using multiple different voices, you will need to run the same pieces of audio through the various profiles in order to get a complete transcription.
2) Scribe only can have a maximum of 6 profiles. This means that you will have to delete and reload your profile every time you work.
1) Make a profile for each member of the group.
2) Make sure you record in the correct file format. MacSpeech Scribe does not except MP3 files.
3) MacSpeech Scribe suggests that when using a digital recorder (like the H4n) it is best to record at the highest possible quality (for the H4n WAV 96kHz/24bit). Keep in mind when recording in this file size, the amount you are able to record decreases significantly.