denote-say

denote-say provides functions for integrating denote notes and any tts engine that can create audio files.

denote-say provides functionality for two kinds of tasks.

Create text files that are suitable for use with a tts engine. This primarily involves removing text that won't be easily voiced by a tts engine, particularly in org-mode files. Examples of problematic text:
- Links.
- Source code blocks.
- Text with emphasis markup.
- Headers.
A text file can be created by combining multiple denote files (files from denote links and dblock links).
Create audio files from those text files (using a tts engine) in a temporary directory and then play them.

Installation

Manual

denote-say is not available on ELPA. To install manually, download denote-say.el, then call M-x package-install-file on it.

M-x package-initialize may be required to recognize the package after installation (just once after installation).

Requirements

Denote 2.0.0 or above is required and any tts engine that can create audio files.

Configuration

The following variables can be use for configuring denote-say.

denote-say-temp-directory

This is where text and audio files are stored. The default value is in the subdirectory denote-say of user-emacs-directory.

denote-say-tts-commands

The list of available tts commands. These commands create an audio file using a text file. The placeholders textfile and audiofile (enclosed in < >) must be used when describing a command.

(defvar denote-say-tts-commands
  `((festival . "text2wave -o <audiofile> <textfile>")
    (festival-es . "text2wave -o <audiofile> -eval '(voice_el_diphone) <textfile>")
    (piper . "cat <textfile> | piper --model  en_US-lessac-medium.onnx --length_scale 1.4 --output_file <audiofile> ")
    (piper-fast . "cat <textfile> | piper --model en_US-lessac-medium.onnx --length_scale 0.8 --output_file <audiofile> ")))

denote-say-tts-command

This is the command used to create audio files. The value is a key of the denote-say-tts-commands alist. The default value is piper.

denote-say-org-replacements

Is the list of replacements that are applied to denote files. The order of the replacements is important; top entries are applied first.

Some replacements involve adding a '.' character; the output sounds better in my opinion.

(defvar denote-say-org-replacements
  `(("^#[+]title: *\\(.*\\)" .                    "\\1." )
    ("\\(^#[+]date:.*\\|^#[+]filetags: *\\|^#[+]identifier:.*\\)" . "")
    ("^#[+]filetags: *:\\(.*\\):" . "Keywords. \\1.")
    (org-babel-src-block-regexp . "Source Code. \\2")
    (org-block-regexp . "D Block Links. \\4.")
    (org-heading-regexp . "\\2.")
    (org-any-link-re . "\\3.")
    (org-emph-re  . "\\4")
    (org-verbatim-re  . " \\4 ")
    (org-target-regexp  . "\\1")
    ("\""  . "")
    ("\\(<<\\|>>\\)"  . "")
    ("\\(;\\|:\\)"  . ".")))

denote-say-play-function

The function used for playing audio files. The default value is emms-play-file.

denote-say-ocr-commands

The list of available OCR commands.

(defvar denote-say-ocr-commands
  `((tesseract . "tesseract <imagefile> <textfile> -l eng")
	(tesseract-spa . "tesseract <imagefile> <textfile> -l spa")))

denote-say-ocr-command

This is the command used for OCR. The value must be a key of the denote-say-ocr-commands alist. The default value is tesseract.

Usage

denote-say-buffer

Creates a text file from the current buffer (using all the replacements form denote-say-org-replacements), then an audio file using denote-say-tts-command, then it plays the audio file using denote-say-play-function.

denote-say-buffer-choose-tts

It does the same as denote-say-buffer but the tts command is selected from a prompt.

denote-say-set-tts-command

Can be used to change the default tts command.

denote-say-find-note

Choose a note from denote-directory and call denote-say-buffer on that file.

denote-say-find-note-choose-tts

Like denote-say-find-note but allows you to choose the tts command.

PDF commands

pdf-tools and an ocr program is required for the following commands.

denote-say-pdf-ocr-page

Use an OCR command on the current PDF page, and then play the resulting text file.

MirkoHernandez/denote-say