Text-to-Speech Latency
The Text-to-Speech Latency report shows response times for speech synthesis requests over the selected period of time. Latency is measured as the time from receipt of the playback request at the API server to the time when the server delivers the initial packet of audio back to the calling application.
In addition to the use of report filters, Nuance Insights allows you to manipulate displayed data through several other means in order to better visualize information. Select from the following to learn more:
Note: This report is applicable only for clients using the Conversational AI channel.
Aggregate data includes the number of requests, outcomes, min, max, and average latency, for each application.
The report only applies to successful requests. Latency is expressed in seconds, for example 0.13.
| Column | Description |
|---|---|
| Web App | Web application |
| API |
API command (for example, |
| TTS Voice | Voice used for speech; a name (for example Carol) |
| Total Requests | Total number of requests for this combination of dimensions |
| Average Playback | Average latency |
| Min Playback | Minimum latency |
| Max Playback | Maximum latency |
Visualizations
The API / TTS histogram graphically represents the number of requests for speech synthesis, grouped by TTS voice which is graphically represented in the histogram through color-coding. The Summary table immediately following the histogram is a 1:1 tabular representation of the same data, on a per voice basis. The Details table at the bottom of the page draws from the same data set but groups the information by date stamp.
Mouse over any of the histogram bars to display a tool tip listing the underlying data for that particular voice.
Filters
Time Range
The Time Range slider allows you to narrow or expand the date range. The data displayed in all visualizations automatically reflect the subset of information falling within the selected range.
Tenant
By choosing one or more tenants from the list, you refine your displayed data-set by including data from only those tenants matching the selection. The visualizations automatically reflect the information complying with the tenant selection.
