tts-callsummary payload
Current version: v3
Nuance Vocalizer for Cloud (NVC) offers two types of synthesis response: streamed and unary.
The synthesized audio is tyypically streamed back to the application (Synthesizer/Synthesize), but a unary (non-streamed, single package) response may be requested instead (Synthesizer/UnarySynthesize). The same payload is returned with the exception of the source
field.
For more information on streamed versus unary responses, see the TTSaaS gRPC API documentation.
Note:
The GetVoice RPC method in the Synthesizer service does not create any event log entries, as it is a purely informational request.In addition to the standard fields described in data field structure, messages with the application/x-nuance-tts-callsummary.v3+json
dataContentType include service-specific fields as detailed below.
Request object
The request
object describes the synthesis resources used to tune the synthesized audio, as provided in the Input message of a SynthesisRequest. Default is blank (empty array).
Resources may include:
- User dictionaries, which provide custom pronunciations
- Rulesets, which apply search-and-replace rules to input text
- ActivePrompt databases, which help tune synthesized audio under certain conditions, using Nuance Vocalizer Studio.
Response object
The response
object, in addition to providing a status
message, provides a list (array) of events. For more information about the synthesis response, see SynthesisResponse.
For a successful response, response.events
will always contain:
-
One or many NVOCcntv events (one per voice used in the request): Number of Unicode input characters for a voice and voice operating point combination processed in a single speak request.
-
One NVOCcntg event (total character count): Number of Unicode input characters processed in a single speak request.
-
One NVOCinpt event: Input text for the speak request.
-
One NVOCsynd event: Logged at the end of synthesis.
Each event includes the following two fields:
-
TTSTIME
, the time the event was produced by the core engine. -
TTSAASTIME
, the time NVC received the event from the core engine. May be used to diagnose slowness between the core engine and NVC.
Sample payload
NVOCcntv event
The NVOCcntv event indicates the number of Unicode input characters processed for a specific voice operating point in a single speak request. One event is logged per voice used in the request.
Field | Description |
---|---|
EVENT | NVOCcntv |
VOICE_VOP | Name of the voice and its operating point |
CHARS | Number of Unicode characters processed |
VOIC | Voice name |
VMDL | Voice model. For example, xpremium (standard) xpremium-high (enhanced). Other values will be returned for legacy voices (full_vssqf22 for the Allison voice, for example). |
LOCALE | Locale ID |
DURS | Output duration in milliseconds |
NVOCcntg event
The NVOCcntg event indicates the total Unicode character count processed in a single speak request.
Field | Description |
---|---|
EVENT | NVOCcntg |
CHARS | Number of Unicode characters processed |
NVOCinpt event
The NVOCinpt event provides the input text for the speak request and is logged at the end of each speak request. If an error occurred or the synthesis request was stopped, NVOCinpt reports the information obtained prior to the error or interruption.
Field | Description |
---|---|
EVENT | NVOCinpt |
MIME | MIME content type for the input text |
TXSZ | Input size in bytes, the raw input prior to any processing |
TEXT | Input text. Omitted when the suppress_input field is true in the SynthesisRequest (default is false). |
NVOCsynd event
The NVOCsynd event is logged at the end of synthesis. If an error occurred or the synthesis request was stopped, NVOCsynd reports the information obtained prior to the error or interruption.
Field | Description |
---|---|
EVENT | NVOCsynd |
INPT | Input size in bytes; the raw input prior to any processing |
DURS | Output duration in milliseconds |
RSTT | Result status:
|
NVOCifst event
The NVOCifst event indicates an Internet fetch has started.
Field | Description |
---|---|
EVENT | NVOCifst |
PROP | Semicolon-separated list of Internet fetch properties, where each property is of the form property=value |
URI | URI being fetched; can be a relative URI |
NVOCifnd event
The NVOCifnd indicates an Internet fetch has ended.
Field | Description |
---|---|
EVENT | NVOCifnd |
FRST | Fetch result: SUCCESS if the fetch succeeded; otherwise a brief error description |
DSRC | Data source:
|
SIZE | Size in bytes. 0 if the fetch failed. |
MIME | MIME content type. Empty string if the fetch failed. |
URI | URI that was fetched. If the fetch was successful, this will be an absolute URI. |
NVOCrlst event
The NVOCrlst event is written when a ruleset is used during synthesis.
Field | Description |
---|---|
EVENT | NVOCrlst |
RULSET_NAME | URI that was fetched for the ruleset. Example:urn:nuance-mix:tag:model/ruleset/mix.tts?=language=en-US |
NVOCdict event
The NVOCdict event is written when a user dictionary is used during synthesis.
Field | Description |
---|---|
EVENT | NVOCdict |
URI | URI that was fetched for the user dictionary. Example:urn:nuance-mix:tag:model/user_dictionary/mix.tts?=language=en-US |
NVOCactp event
The NVOCactp event is written when an ActivePrompt is used during synthesis. These events are written in the order in which the ActivePrompts are matched (highest to lowest precedence across the input text), not the order in which they appear in the output audio stream.
Field | Description |
---|---|
EVENT | NVOCactp |
APID | ActivePrompt ID of the form domain::prompt |
NVOCadon event
The NVOCadon event is written when a system-level dictionary is used during synthesis.
Field | Description |
---|---|
EVENT | NVOCadon |
ADDON_NAME | URI that was fetched for the add-on. Example:urn:nuance-mix:tag:model/system_dictionary/mix.tts?=language=en-US |
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.