Configuration of the Speech to Text feature consists of the following steps, detailed below:
Obtain a Nuance speech transcription account and conversion credits
Configure system-wide Speech to Text settings
Enable the Speech to Text feature
Configure Speech to Text for users
Nuance Communications is responsible for transcribing voice messages to text. To access this service, you require a user account and conversion credits, which you can obtain through your Authorized Mitel Reseller.
Note: Purchase the NuPoint UM Speech to Text "enablement" license before you set up your Nuance account and obtain conversion credits.
Contact your Authorized Mitel Reseller and provide the following:
User Name: Name of your NuPoint UM system.
System ID: Application Record ID, or ARID, of the NuPoint UM system.
Conversion credit quantity: Mitel sells conversion credits in prepackaged quantities.
Confirmation email addressess: Nuance sends a confirmation email to these addresses when conversion credits are purchased.
The information is sent to Nuance, which sets up your user account, assigns conversion credits, and sends you an email containing the following:
User Name: Same as above.
Password: Your unique account password, provided by Nuance.
Account ID: Your unique account ID number, provided by Nuance.
Application ID: Your unique application ID name, provided by Nuance.
Complete the next procedure, Configuring System-wide Speech to Text Settings.
These same credentials are used for account renewals. NuPoint UM does not know or report the state of your Nuance subscription. If your account is running low, Nuance notifies the NuPoint UM server, which in turn issues warning alarm messages. If your account is empty, the Speech to Text transcription service is terminated. You will need to purchase additional credits through your Authorized Mitel Reseller and then Activate Text to Speech.
In the NuPoint UM Web Console:
From the navigation tree, click Unified Messaging > STT Configuration. If properly licensed, the Speech to Text configuration page is displayed.
Enter the following:
Username: Name of your NuPoint UM system.
Password: Your unique account password, provided by Nuance.
Account ID: Your unique account ID number, provided by Nuance.
Application ID: Your unique application ID name, provided by Nuance.
To include the original audio attachment of the voice message in the transcription emails for all users, select Include audio attachments with email messages.
Click Save.
Customize an FCOS to include the following feature bits:
285 (Enable Speech to Text) and assign it to mailboxes that will use this feature.
295 (and an Advanced UM license) OR 304 (and a UM Standard license) OR 289 (and a UM-SMTP license).
290 to view and/or save a text transcription of a voice mail message in the Web View.
Customize an LCOS with a Minimum Message Length greater than zero (0).
Note: If you fail to set a minimum message length, brief messages (such as when a caller immediately hangs up) will be transcribed and cause a conversion credit to be consumed.
Apply FCOS and LCOS to the STT mailboxes.
Notes:
To include the original audio attachment of the voice message with the text transcription email for all users, enable the Include Audio Attachments system-wide settings.
All users must accept the End User License Agreement before they can use the Speech to Text feature. (Administrators cannot accept on behalf of users.)
The appropriate feature bits must be assigned to the FCOS of the user's mailbox before the Speech to Text configuration fields are displayed in the Web Console.
Users should record a voice mail greeting to encourage callers to speak clearly. For example: "Please speak clearly as your voice message will be transcribed and sent to me in an email."
To configure STT for users:
From the navigation tree, click Mailbox Maintenance, and then click Mailboxes.
Search for a specific mailbox or click Show All to see a complete list of mailboxes.
Select a mailbox in the list, and then click Edit > Selected. The Mailbox data view is displayed (Basic view), populated with data for the selected mailbox.
For UM-SMTP Users:
Enter a valid email address in the UM-SMTP Email Address field.
From the delivery option list beside the Email Address field, select Speech-to-Text. All incoming voice mail messages (except Confidential, Record-a-Call, and Fax messages) are transcribed into text and sent as email messages to the specified email address.
Click Save.
For UM Standard Users:
Enter a valid email address in at least one of the Standard UM Email Address fields.
From the delivery option list beside the appropriate Email Address field, select Speech-to-Text. All incoming voice mail messages (except Confidential, Record-a-Call, and Fax messages) are transcribed into text and sent as separate email messages to the specified email address.
Click Save.
Note: For UM Standard Users, it is possible to configure Speech to Text using the Text Console (see the NuPoint Unified Messaging System Administration online help, available at Mitel OnLine).
For UM Advanced Users:
Note: Prior to using STT for UM Advanced users, you must program the smart host in the MSL Server Manager on the E-Mail settings screen. The smart host can be entered as IP address or a FQDN.
Select the Enable Speech-to-Text Transcription check box. All incoming voice messages are automatically transcribed and sent as email messages to the user's account. Note: The transcriptions are sent as separate emails and are not synchronized. (Only the original voice message with audio attachment is synchronized with the user's Inbox and/or MWI.)
Click Save.
Note: To take advantage of Secure IMAP, you must enter the correct authentication settings in the Exchange Server.
The STT feature is not a dictation service. While transcription accuracy can be very good, there may be instances when the transcription does not accurately represent the spoken message. If in doubt, listen to the original voice message.
Most callers do not speak in full and complete sentences; the transcription service will attempt to reflect this with spaces and punctuation.
Transcription quality depends on the clarity of the original voice message. For example, if the person has a heavy accent or does not speak clearly, or is speaking from a noisy from a noisy environment, then the message will not be transcribed correctly.
If the system cannot understand a voice message, it will not transcribe it. Instead, the user will receive an email notification. Typically, about 15% of messages are deemed "untranscribable."
The system is "tuned" to transcribe typical English conversation. In many businesses however, messages may contain jargon, phrases and acronyms that are difficult to transcribe.
The service is available in North American English only. For this reason, users who receive a significant number of voice messages in a different language should not be enabled for STT.
The maximum message length is 60 seconds. For longer messages, users can dial in and listen to the entire original voice mail.