SprexTM Home
TallyGramTM
ANSRTM
TeachionaryTM
SprexPassTM Passwords
PhonolyzeTM
Lip-Synch MachineTM
Consulting Services
SprexOutTM TTS
Co-WorkTM
Exit
phono [sound] # lyze [cut] |
Background
Segmentation of audio into phones or words is a highly time-consuming and painstaking process. Required for a variety of applications, such as acoustic analysis studies in academia, or building speech synthesis databases for high-quality speech output systems, paid staff time to carry out this process for even relatively small audio databases can rapidly accumulate to prohibitive expenditure levels. (although the rare graduate student might perhaps be inspired to do this work for no cost.)By using the PhonolyzeTM system to automatically segment audio into phones and (if a transcript is available) words, much of this painstaking work can be bypassed, reducing to the faster task of merely checking and doing minor corrections, depending on the level of precision required.
Documentation
Please read the manual page which documents the command-line interface to the phonolyze program.Requirements
- SNR and Language: For
PhonolyzeTM
to achieve optimum accuracy, signal to noise levels must
be adequate (30dB), and the speech must be from a supported language
(US English comes preinstalled, while add-on
modules for Spanish, UK English, German, and Japanese are also
available). If you have a different language to process, it's worth a
try; sometimes results are better than the alternative.
- Platform: Phonolyze is supported on Mandrake Linux 8.x, FreeBSD 4.x, Solaris 4.1.3, and Windows 2000. Support for other versions of these platforms is intended and likely but not guaranteed.
What next
Please read our on-line documentation above. If you would like a PhonolyzeTM demonstration using your own audio data, preparatory to a purchase of the product, please contact us to discuss your application and request a demonstration and a quotation. For the demonstration, we will ask you for a tarball of audio samples (16000Hz 16-bit linear PCM audio in raw or .wav format) and we'll generate phone- and/or word-level segmentations or transcriptions for you based on your data, so that you can compare it against independent segmentation of the same data.