I use the Rockbox software on my Cowon iAudio X5L player for several years and now on Sansa Clip+ as well with great pleasure. During this time I occasionally tried to improve it’s functionality in some aspects to make it yet better fitting my needs. The results of these activities are represented here.
See the project homepage for official Rockbox releases, accompanying documentation and other useful resources related to this excellent software, but here you can find only my apocryphal builds, information about the applied patches and some additional stuff.
Release date: 2020-12-27.
These builds are based on the master branch of the Rockbox development source and include a bunch of fixes and enhancements as well as the latest main project updates up to build date. Summary of the master changes history is available here.
All fonts and English voice generated by Mbrola speech synthesizer are included. Being downloaded and properly installed on a device all this should work just out of the box. In addition, auto-generated Russian voice is provided for each build as a separate archive. Simply unpack it alongside the respective Rockbox build if you need.
Choose an appropriate build for your architecture:
In all these builds speech feedback initially is turned on only for
menu. But some menu sections, such as font or language choice, in
fact, are implemented as file lists. To make them speaking as well, it
is necessary to go to
General settings /
Voice directories and
Voice file names and enable using talk
clips both for directories and files.
Since personally I use Sansa Clip+ and Cowon iAudio X5, all fixes and additions are tested by me only on these architectures. Other builds were created for my friends who like these improvements and asked me to do that.
Since the official Rockbox bootloader for Cowon iAudio X5 player overrides the original firmware, I prefer the alternative one that allows to keep Rockbox along with the original firmware. See the README file included into the archive for details.
Although Rockbox natively supports a bunch of sound file formats, but
if you want it to be able to play midi-files as well, grab
and unpack it in the
.rockbox folder on your player after
Rockbox installation. It should be enough.
The voicebox shell script is designed to make voice thumbnails for file names and FM radio presets on a Linux machine. It makes use of the rbspeak shell script for actual speech synthesis. This script in turn uses Mbrola speech synthesizer for producing English speech and Ru_tts engine for Russian. See embedded comments for more details. To compress produced sound the rbspeexenc program is used that was generated from the Rockbox sources.
On Debian and Ubuntu all necessary speech synthesizers can be installed from my repository by command:
$ sudo aptitude install mbrola mbrola-en1 freephone enlex ru-tts rulex
An extra advantage of this tool is that it uses different voice pitch for different objects, thus, folders and files of various types are distinguishable simply by voice.
The FM radio presets pack for a bunch of cities (mostly Russian) along
with voice thumbnails is available here.
Downloaded archive should be unpacked in the
.rockbox directory on
the device where Rockbox is installed. The voice thumbnails were
created by Russian and English speech synthesizers depending on the
alphabet used in the station names. The base information about radio
stations and their frequencies was taken from
In order to simplify switching from one source to another in recorder I’ve created basic presets for each one:
/voicedirectory, 22 kHz mono compressed as mp3 64kbps;
/radiodirectory, 44 kHz stereo compressed as mp3 128 kbps;
/recorddirectory, 44 kHz stereo PCM wave format.
To make use of these presets grab the archive and unpack it into
/.rockbox folder on the device.
Feel free to e-mail me with all related questions and suggestions.