gnu/packages/speech.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2016 David Thompson <davet@gnu.org>
   3 ;;; Copyright © 2016, 2019 Marius Bakke <mbakke@fastmail.com>
   4 ;;; Copyright © 2017 Leo Famulari <leo@famulari.name>
   5 ;;; Copyright © 2018 Tobias Geerinckx-Rice <me@tobias.gr>
   6 ;;; Copyright © 2016 Kei Kebreau <kkebreau@posteo.net>
   7 ;;; Copyright © 2019 Ricardo Wurmus <rekado@elephly.net>
   8 ;;;
   9 ;;; This file is part of GNU Guix.
  10 ;;;
  11 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  12 ;;; under the terms of the GNU General Public License as published by
  13 ;;; the Free Software Foundation; either version 3 of the License, or (at
  14 ;;; your option) any later version.
  15 ;;;
  16 ;;; GNU Guix is distributed in the hope that it will be useful, but
  17 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;;; GNU General Public License for more details.
  20 ;;;
  21 ;;; You should have received a copy of the GNU General Public License
  22 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  23
  24 (define-module (gnu packages speech)
  25   #:use-module ((guix licenses) #:prefix license:)
  26   #:use-module (guix packages)
  27   #:use-module (guix download)
  28   #:use-module (guix git-download)
  29   #:use-module (guix utils)
  30   #:use-module (guix build-system gnu)
  31   #:use-module (gnu packages)
  32   #:use-module (gnu packages audio)
  33   #:use-module (gnu packages autotools)
  34   #:use-module (gnu packages compression)
  35   #:use-module (gnu packages emacs)
  36   #:use-module (gnu packages gcc)
  37   #:use-module (gnu packages glib)
  38   #:use-module (gnu packages linux)
  39   #:use-module (gnu packages ncurses)
  40   #:use-module (gnu packages pkg-config)
  41   #:use-module (gnu packages pulseaudio)
  42   #:use-module (gnu packages python)
  43   #:use-module (gnu packages texinfo)
  44   #:use-module (gnu packages textutils))
  45
  46 (define-public espeak
  47   (package
  48     (name "espeak")
  49     (version "1.48.04")
  50     (source (origin
  51               (method url-fetch)
  52               (uri (string-append "mirror://sourceforge/espeak/espeak/"
  53                                   "espeak-" (version-major+minor version)
  54                                   "/espeak-" version "-source.zip"))
  55               (sha256
  56                (base32
  57                 "0n86gwh9pw0jqqpdz7mxggllfr8k0r7pc67ayy7w5z6z79kig6mz"))
  58               (modules '((guix build utils)))
  59               (snippet
  60                ;; remove prebuilt binaries
  61                '(begin
  62                   (delete-file-recursively "linux_32bit")
  63                   #t))))
  64     (build-system gnu-build-system)
  65     (arguments
  66      `(#:make-flags (list (string-append "PREFIX=" (assoc-ref %outputs "out"))
  67                           (string-append "DATADIR="
  68                                          (assoc-ref %outputs "out")
  69                                          "/share/espeak-data")
  70                           (string-append "LDFLAGS=-Wl,-rpath="
  71                                          (assoc-ref %outputs "out")
  72                                          "/lib")
  73                           ;; The package fails to build with newer C++ standards.
  74                           "CXXFLAGS=-std=c++98"
  75                           "AUDIO=pulseaudio")
  76        #:tests? #f ; no check target
  77        #:phases
  78        (modify-phases %standard-phases
  79          (replace 'configure
  80            (lambda _
  81              (chdir "src")
  82              ;; We use version 19 of the PortAudio library, so we must copy the
  83              ;; corresponding file to be sure that espeak compiles correctly.
  84              (copy-file "portaudio19.h" "portaudio.h")
  85              (substitute* "Makefile"
  86                (("/bin/ln") "ln"))
  87              #t)))))
  88        (inputs
  89         `(("portaudio" ,portaudio)
  90           ("pulseaudio" ,pulseaudio)))
  91        (native-inputs `(("unzip" ,unzip)))
  92        (home-page "http://espeak.sourceforge.net/")
  93        (synopsis "Software speech synthesizer")
  94        (description "eSpeak is a software speech synthesizer for English and
  95 other languages.  eSpeak uses a \"formant synthesis\" method.  This allows many
  96 languages to be provided in a small size.  The speech is clear, and can be used
  97 at high speeds, but is not as natural or smooth as larger synthesizers which are
  98 based on human speech recordings.")
  99        (license license:gpl3+)))
 100
 101 (define-public espeak-ng
 102   (package
 103     (name "espeak-ng")
 104     (version "1.49.2")
 105     (home-page "https://github.com/espeak-ng/espeak-ng")
 106     (source (origin
 107               (method url-fetch)
 108               (uri (string-append home-page "/releases/download/" version
 109                                   "/espeak-ng-" version ".tar.gz"))
 110               (sha256
 111                (base32 "1d10x9rbvqi2zwcz65fxh04k0x0scnk7732l37laz6xra1ldhzng"))))
 112     (build-system gnu-build-system)
 113     (arguments
 114      `(#:configure-flags '("--disable-static")
 115        ;; Building in parallel triggers a race condition in 1.49.2.
 116        #:parallel-build? #f
 117        ;; XXX: Some tests require an audio device.
 118        #:tests? #f))
 119     (inputs
 120      `(("libcap" ,libcap)
 121        ("pcaudiolib" ,pcaudiolib)))
 122     (synopsis "Software speech synthesizer")
 123     (description
 124      "eSpeak NG is a software speech synthesizer for more than 100 languages.
 125 It is based on the eSpeak engine and supports spectral and Klatt formant
 126 synthesis, and the ability to use MBROLA voices.")
 127     (license license:gpl3+)))
 128
 129 (define-public mitlm
 130   (package
 131     (name "mitlm")
 132     (version "0.4.2")
 133     (source (origin
 134               (method url-fetch)
 135               (uri (string-append "https://github.com/mitlm/mitlm/releases/"
 136                                   "download/v" version "/"
 137                                   name "-" version ".tar.xz"))
 138               (sha256
 139                (base32
 140                 "09fv4fcpmw9g1j0zml0k5kk1lgjw2spr8gn51llbkaaph6v8d62a"))))
 141     (build-system gnu-build-system)
 142     (native-inputs
 143      `(("gfortran" ,gfortran)))
 144     (synopsis "The MIT Language Modeling toolkit")
 145     (description "The MIT Language Modeling (MITLM) toolkit is a set of
 146 tools designed for the efficient estimation of statistical n-gram language
 147 models involving iterative parameter estimation.  It achieves much of its
 148 efficiency through the use of a compact vector representation of n-grams.")
 149     (home-page "https://github.com/mitlm/mitlm")
 150     (license license:expat)))
 151
 152 (define-public speech-dispatcher
 153   (package
 154     (name "speech-dispatcher")
 155     (version "0.9.1")
 156     (source (origin
 157               (method url-fetch)
 158               (uri (string-append "https://github.com/brailcom/speechd/releases"
 159                                   "/download/" version "/speech-dispatcher-"
 160                                   version ".tar.gz"))
 161               (sha256
 162                (base32
 163                 "16bg52hnkrsrs7kgbzanb34b9zb6fqxwj0a9bmsxmj1skkil1h1p"))))
 164     (build-system gnu-build-system)
 165     (arguments
 166      `(#:configure-flags '("--disable-static"
 167
 168                            ;; Disable support for proprietary TTS engines.
 169                            "--with-ibmtts=no"
 170                            "--with-kali=no" "--with-baratinoo=no")))
 171     (native-inputs
 172      `(("intltool" ,intltool)
 173        ("pkg-config" ,pkg-config)))
 174     (inputs
 175      `(("dotconf" ,dotconf)
 176        ("espeak" ,espeak-ng)
 177        ("glib" ,glib)
 178        ("libltdl" ,libltdl)
 179        ("libsndfile" ,libsndfile)
 180        ("pulseaudio" ,pulseaudio)
 181        ("python" ,python)))
 182     (synopsis "Common interface to speech synthesizers")
 183     (description "The Speech Dispatcher project provides a high-level
 184 device independent layer for access to speech synthesis through a simple,
 185 stable and well documented interface.")
 186     (home-page "https://devel.freebsoft.org/speechd")
 187     ;; The software is distributed under GPL2+, but includes a number
 188     ;; of files covered by other licenses.  Note: in practice, this
 189     ;; is linked against dotconf, which is LGPL 2.1 only.
 190     (license (list license:gpl2+
 191                    license:fdl1.2+ ; Most files in doc/ are dual gpl2+/fdl1.2+.
 192                    license:lgpl2.1+
 193                    (license:non-copyleft
 194                     ;; festival_client.{c,h} carries an expat-style license.
 195                     "See src/modules/festival_client.c in the distribution.")
 196                    license:gpl3+)))) ; doc/texinfo.tex -- with TeX exception.
 197
 198 (define-public sonic
 199   (package
 200     (name "sonic")
 201     (version "0.2.0")
 202     (source (origin
 203              (method git-fetch)
 204              (uri (git-reference
 205                     (url "https://github.com/waywardgeek/sonic")
 206                     (commit (string-append "release-" version))))
 207              (file-name (git-file-name name version))
 208              (sha256
 209               (base32
 210                "08xwnpw9cnaix1n1i7gvpq5hrfrqc2z1snjhjapfam506hrc77g4"))))
 211     (build-system gnu-build-system)
 212     (arguments
 213       `(#:tests? #f ; No test suite.
 214         #:make-flags
 215          (list (string-append "DESTDIR=" (assoc-ref %outputs "out")))
 216         #:phases
 217         (modify-phases %standard-phases
 218           (delete 'configure)))) ; No ./configure script.
 219     (synopsis "Speed up or slow down speech")
 220     (description "Sonic implements a simple algorithm for speeding up or slowing
 221 down speech.  However, it's optimized for speed ups of over 2X, unlike previous
 222 algorithms for changing speech rate.  Sonic is a C library designed to be easily
 223 integrated into streaming voice applications such as text-to-speech (TTS) back
 224 ends.
 225
 226 The primary motivation behind Sonic is to enable the blind and visually impaired
 227 to improve their productivity with speech engines, like eSpeak.  Sonic can also
 228 be used by the sighted.")
 229     (home-page "https://github.com/waywardgeek/sonic")
 230     (license license:asl2.0)))
 231
 232 (define-public festival
 233   (package
 234     (name "festival")
 235     (version "2.5.0")
 236     (source (origin
 237               (method url-fetch)
 238               (uri (string-append "http://festvox.org/packed/festival/"
 239                                   (version-major+minor version)
 240                                   "/festival-" version "-release.tar.gz"))
 241               (sha256
 242                (base32
 243                 "1d5415nckiv19adxisxfm1l1xxfyw88g87ckkmcr0lhjdd10g42c"))))
 244     (build-system gnu-build-system)
 245     (arguments
 246      `(#:tests? #f ; there is no test target
 247        #:make-flags
 248        (list (string-append "RM="
 249                             (assoc-ref %build-inputs "coreutils")
 250                             "/bin/rm")
 251              (string-append "ECHO_N="
 252                             (assoc-ref %build-inputs "coreutils")
 253                             "/bin/printf \"%s\""))
 254        #:parallel-build? #f ; not supported
 255        #:modules ((guix build gnu-build-system)
 256                   (guix build utils)
 257                   (guix build emacs-utils))
 258        #:imported-modules (,@%gnu-build-system-modules
 259                            (guix build emacs-utils))
 260        #:phases
 261        (modify-phases %standard-phases
 262          (add-after 'unpack 'unpack-and-patch-speech-tools
 263            (lambda* (#:key inputs #:allow-other-keys)
 264              (invoke "tar" "-C" ".."
 265                      "-xf" (assoc-ref inputs "speech-tools"))
 266              (with-directory-excursion "../speech_tools"
 267                (substitute* '("config/rules/modules.mak"
 268                               "config/rules/test_make_rules.mak"
 269                               "config/make_system.mak")
 270                  (("/bin/sh") (which "sh"))))
 271              #t))
 272          (add-after 'unpack 'patch-/bin/sh
 273            (lambda _
 274              (substitute* '("config/test_make_rules"
 275                             "config/make_system.mak")
 276                (("/bin/sh") (which "sh")))
 277              #t))
 278          (add-before 'build 'build-speech-tools
 279            (lambda* (#:key configure-flags make-flags #:allow-other-keys)
 280              (with-directory-excursion "../speech_tools"
 281                (apply invoke "sh" "configure"
 282                       (string-append "CONFIG_SHELL=" (which "sh"))
 283                       (string-append "SHELL=" (which "sh"))
 284                       configure-flags)
 285                (apply invoke "make" make-flags))))
 286          (add-after 'build 'build-documentation
 287            (lambda _
 288              (with-directory-excursion "doc"
 289                (invoke "make" "festival.info"))))
 290          (add-after 'unpack 'set-installation-directories
 291            (lambda* (#:key outputs #:allow-other-keys)
 292              (let ((out (assoc-ref outputs "out")))
 293                (substitute* "config/project.mak"
 294                  (("^FTLIBDIR.*")
 295                   (string-append "FTLIBDIR=" out "/share/festival/lib")))
 296                (substitute* "config/systems/default.mak"
 297                  (("^INSTALL_PREFIX.*")
 298                   (string-append "INSTALL_PREFIX=" out)))
 299                #t)))
 300          (add-after 'install 'actually-install
 301            (lambda* (#:key inputs outputs #:allow-other-keys)
 302              (let ((out (assoc-ref outputs "out")))
 303                ;; Install Speech Tools first
 304                (with-directory-excursion "../speech_tools"
 305                  ;; Target directories
 306                  (for-each (lambda (dir)
 307                              (mkdir-p (string-append out dir)))
 308                            '("/bin"
 309                              "/lib"
 310                              "/include/speech_tools/"
 311                              "/include/speech_tools/instantiate"
 312                              "/include/speech_tools/ling_class"
 313                              "/include/speech_tools/rxp"
 314                              "/include/speech_tools/sigpr"
 315                              "/include/speech_tools/unix"))
 316                  ;; Install binaries
 317                  (for-each (lambda (file)
 318                              (install-file file (string-append out "/bin")))
 319                            (find-files "bin" ".*"))
 320                  (for-each (lambda (file)
 321                              (delete-file (string-append out "/bin/" file)))
 322                            '("est_gdb" "est_examples" "est_program"))
 323                  ;; Install libraries
 324                  (for-each (lambda (file)
 325                              (install-file file (string-append out "/lib")))
 326                            (find-files "lib" "lib.*\\.so.*"))
 327
 328                  ;; Install headers
 329                  (for-each
 330                   (lambda (dir)
 331                     (for-each
 332                      (lambda (header)
 333                        (install-file header
 334                                      (string-append out "/include/speech_tools/" dir)))
 335                      (find-files (string-append "include/" dir)
 336                                  "\\.h$")))
 337                   '("." "instantiate" "ling_class" "rxp" "sigpr" "unix")))
 338
 339                ;; Unpack files that will be installed together with the
 340                ;; Festival libraries.
 341                (invoke "tar" "--strip-components=1"
 342                        "-xvf" (assoc-ref inputs "festvox-cmu"))
 343                (invoke "tar" "--strip-components=1"
 344                        "-xvf" (assoc-ref inputs "festvox-poslex"))
 345                (invoke "tar" "--strip-components=1"
 346                        "-xvf" (assoc-ref inputs "default-voice"))
 347
 348                ;; Install Festival
 349                (let ((bin (string-append out "/bin"))
 350                      (incdir (string-append out "/include/festival"))
 351                      (share (string-append out "/share/festival"))
 352                      (info (string-append out "/share/info")))
 353                  (for-each (lambda (executable)
 354                              (install-file executable bin))
 355                            '("src/main/festival"
 356                              "src/main/festival_client"
 357                              "examples/benchmark"))
 358                  (let ((scripts '("examples/dumpfeats"
 359                                   "examples/durmeanstd"
 360                                   "examples/latest"
 361                                   "examples/make_utts"
 362                                   "examples/powmeanstd"
 363                                   "examples/run-festival-script"
 364                                   "examples/saytime"
 365                                   "examples/scfg_parse_text"
 366                                   "examples/text2pos"
 367                                   "examples/text2wave")))
 368                    (substitute* scripts
 369                      (("exec /tmp/guix-build.*/bin/festival")
 370                       (string-append "exec " bin "/festival")))
 371                    (for-each (lambda (script)
 372                                (install-file script bin))
 373                              scripts))
 374
 375                  ;; Documentation
 376                  (for-each (lambda (file)
 377                              (install-file file info))
 378                            (find-files "doc/info/" "festival.info.*"))
 379
 380                  ;; Headers
 381                  (mkdir-p incdir)
 382                  (for-each (lambda (header)
 383                              (install-file header
 384                                            (string-append incdir "/"
 385                                                           (dirname header))))
 386                            (find-files "src/include" "\\.h$"))
 387
 388                  ;; Data
 389                  (mkdir-p share)
 390                  (for-each (lambda (file)
 391                              (install-file file
 392                                            (string-append share "/"
 393                                                           (dirname file))))
 394                            (find-files "lib" ".*"))
 395                  (for-each delete-file
 396                            (append (find-files share "Makefile")
 397                                    (find-files bin "Makefile")))))
 398              #t))
 399          (add-after 'actually-install 'install-emacs-mode
 400            (lambda* (#:key outputs #:allow-other-keys)
 401              (let ((emacs-dir (string-append (assoc-ref outputs "out")
 402                                              "/share/emacs/site-lisp")))
 403                (install-file "lib/festival.el" emacs-dir)
 404                (emacs-generate-autoloads ,name emacs-dir)
 405                #t)))
 406          ;; Rebuild the very old configure script that is confused by extra
 407          ;; arguments.
 408          (add-before 'configure 'bootstrap
 409            (lambda _ (invoke "autoreconf" "-vif"))))))
 410     (inputs
 411      `(("ncurses" ,ncurses)))
 412     (native-inputs
 413      `(("autoconf" ,autoconf)
 414        ("automake" ,automake)
 415        ("texinfo" ,texinfo)
 416        ("emacs" ,emacs-minimal)
 417        ("festvox-cmu"
 418         ,(origin
 419            (method url-fetch)
 420            (uri (string-append "http://festvox.org/packed/festival/"
 421                                (version-major+minor version)
 422                                "/festlex_CMU.tar.gz"))
 423            (sha256
 424             (base32
 425              "01vwidqhhg2zifvk1gby91mckg1z2pv2mj6lihvdaifakf8k1561"))))
 426        ("festvox-poslex"
 427         ,(origin
 428            (method url-fetch)
 429            (uri (string-append "http://festvox.org/packed/festival/"
 430                                (version-major+minor version)
 431                                "/festlex_POSLEX.tar.gz"))
 432            (sha256
 433             (base32
 434              "18wywilxaqwy63lc47p5g5529mpxhslibh1bjij0snxx5mjf7ip7"))))
 435        ("default-voice"
 436         ,(origin
 437            (method url-fetch)
 438            (uri (string-append "http://festvox.org/packed/festival/"
 439                                (version-major+minor version)
 440                                "/voices/festvox_kallpc16k.tar.gz"))
 441            (sha256
 442             (base32
 443              "136hmsyiwnlg2qwa508dy0imf19mzrb5r3dmb2kg8kcyxnslm740"))))
 444        ("speech-tools"
 445         ,(origin
 446            (method url-fetch)
 447            (uri (string-append "http://festvox.org/packed/festival/"
 448                                (version-major+minor version)
 449                                "/speech_tools-" version "-release.tar.gz"))
 450            (sha256
 451             (base32
 452              "1k2xh13miyv48gh06rgsq2vj25xwj7z6vwq9ilsn8i7ig3nrgzg4"))))))
 453     (home-page "http://www.cstr.ed.ac.uk/projects/festival/")
 454     (synopsis "Speech synthesis system")
 455     (description "Festival offers a general framework for building speech
 456 synthesis systems as well as including examples of various modules.  As a
 457 whole it offers full text to speech through a number APIs: from shell level,
 458 though a Scheme command interpreter, as a C++ library, from Java, and an Emacs
 459 interface.  Festival is multi-lingual though English is the most advanced.
 460 The system is written in C++ and uses the Edinburgh Speech Tools Library for
 461 low level architecture and has a Scheme (SIOD) based command interpreter for
 462 control.")
 463     (license (license:non-copyleft "file://COPYING"))))