Module speech_to_text
Speech to text translation (experimental).
WARNING: totally experimental, no guarantees it will work!
This module provides speech to text translation.
The module requires the Lua lua-cjson and luasec packages.
Info:
- Copyright: 2011-2018 Chad Phillips
- Author: Chad Phillips
Actions
| speech_to_text_from_file | Translates a sound file to text. |
Handlers
| watson | The Watson handler (default). |
Tables
| speech_to_text | Speech to text storage. |
Actions
- speech_to_text_from_file
-
Translates a sound file to text.
This action requires that flac is installed and executable by FreeSWITCH.
- action string speech_to_text_from_file
- api_key string (Optional) The API key used to access the service.
- filepath string The full path to the file to translate.
- storage_area string (Optional) The storage area to store the response in. Defaults to 'speech_to_text'.
Usage:
{ action = "speech_to_text_from_file", api_key = profile.speech_to_text_app_key, filepath = "/tmp/foo.wav", storage_area = "foo_to_text", }
Handlers
- watson
-
The Watson handler (default).
Uses Watson's Speech to Text service. The service requires a valid developer account and api key, see here for more information.
Usage:
{ action = "speech_to_text_from_file", handler = "watson", -- ...other required params... service_uri = "[obtain from service]", -- ...other optional params... query_parameters = { -- key/value pairs to pass as query parameters }, retries = 3, retry_wait_seconds = 60, }
Tables
- speech_to_text
-
Speech to text storage.
Translations are stored in the specified storage area with the following keys, where
Xis the chunk number:A 'status' key is also placed in the storage area, indicating the result of the translation. A value of 0 indicates the translation was successful.
Fields:
- translation_X The translated text for the chunk.
- confidence_X The confidence level of the translated chunk, a decimal number in the range of 0 to 1.