Timing information can also be returned by the API to synchronize speech with other modalities (e.g., text, video) and is available for both standard and streaming requests. Each punctuation sequence and word is associated with a start time and duration (in seconds), which can be used to determine when the text is spoken and how long it lasts. This can be useful for creating captions, subtitles, or aligning speech with other media.

​
Standard Example

This code sample uses the return_durations option to fetch timing information and print out its content.

Code:

import asyncio
from lmnt.api import Speech

async def main():
  options = { 'return_durations': True }
  async with Speech() as speech:
    synthesis = await speech.synthesize('Hello world.', 'lily', **options)
  for chunk in synthesis['durations']:
    print(f'"{chunk["text"]}" starts at {chunk["start"]:.3f}s and lasts for {chunk["duration"]:.3f}s')

asyncio.run(main())

Response:

"" starts at 0.000s and lasts for 0.525s
"Hello" starts at 0.525s and lasts for 0.325s
" " starts at 0.850s and lasts for 0.000s
"world" starts at 0.850s and lasts for 0.400s
"." starts at 1.250s and lasts for 0.375s
"" starts at 1.625s and lasts for 0.013s

​
Streaming example

The code below depicts how to fetch timing information with a streaming request. It is a simple example, and in practice, you would want to set up reader/writer tasks to handle the text input and synthesis output concurrently (see our Streaming example).

Input code:

The option to return timing information in streaming requests is called return_extras. This option name is different from standard requests, where it’s called return_durations.

import asyncio
from lmnt.api import Speech

async def main():
  async with Speech() as speech:
    options = { 'return_extras': True }
    synthesis = await speech.synthesize_streaming('lily', **options)
    await synthesis.append_text('Hello world.')
    await synthesis.finish()
    async for synthesis in synthesis:
      for chunk in synthesis['durations']:
        print(f'"{chunk["text"]}" starts at {chunk["start"]:.3f}s and lasts for {chunk["duration"]:.3f}s')

asyncio.run(main())

Output response:

"" starts at 0.000s and lasts for 0.525s
"Hello" starts at 0.525s and lasts for 0.325s
" " starts at 0.850s and lasts for 0.000s
"world" starts at 0.850s and lasts for 0.400s
"." starts at 1.250s and lasts for 0.375s
"" starts at 1.625s and lasts for 0.013s

Was this page helpful?

Overriding pronunciationsLanguages