[hts-users:02015] Poor voice quality & embedded reestimation (monophone) errors
- Subject: [hts-users:02015] Poor voice quality & embedded reestimation (monophone) errors
- From: Tomasz Kuczmarski <faqster@xxxxxxxxx>
- Date: Mon, 15 Jun 2009 19:29:50 +0200
- Delivered-to: hts-users@xxxxxxxxxxxxxxx
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:date:message-id:subject :from:to:content-type; bh=gHu8Bu/4U8PGQykhkkWmRto7phIz9KaIxgZD6bPD1QE=; b=ZwTWg1KqCRtQVsRZAVcaGNbhaAJd2Qj38s8A8XpWyOeVR3+aPB6uNR9JJKIOr4Z7X2 bKdp9OBWa9q037NcRXun6guZX5NIV5Ohr71AJkhw2gCFER+4v+gXPPGQA3hxzpqZZ7Ju PF9kP8HT6lmWrowXjjWiGGAycnI0Ezpnb30r4=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:date:message-id:subject:from:to:content-type; b=uBxID73jbz26Oh2kEXalKCmr4ai51t1rri6Auuh/NmZSB3Pw4E6DaFJv+RyPT+Gdq4 7G4RxaDQ+8fLWxmYxipk4VdmCz5cZvU0I9xiJ52eu3X0CpAHD1eKL44F0uP7E7wr18uM GUCPt0v5dA8azS/NOmWiF5EFA00ZYu7Zvd8A4=
Hello everybody!
I have just synthesized some test utterances using HMMs trained on a database of Polish speech. The voice quality is sometimes very good but in most of the cases it is unacceptably low - some phones are full of glitches and noise and some of them have incorrect durations.
During the training part of the HTS demo script I got a lot of warnings (for at least 1/3 of all my recordings) during the embedded reestimation (monophone) part:
Processing Data: amu_pl_ilo_A0008.cmp; Label amu_pl_ilo_A0008.lab
Retrying Beta pass at 1600.0
Retrying Beta pass at 1700.0
Retrying Beta pass at 1800.0
Retrying Beta pass at 1900.0
Retrying Beta pass at 2000.0
Retrying Beta pass at 2100.0
Retrying Beta pass at 2200.0
Retrying Beta pass at 2300.0
Retrying Beta pass at 2400.0
Retrying Beta pass at 2500.0
Retrying Beta pass at 2600.0
Retrying Beta pass at 2700.0
Retrying Beta pass at 2800.0
Retrying Beta pass at 2900.0
Retrying Beta pass at 3000.0
Retrying Beta pass at 3100.0
Retrying Beta pass at 3200.0
Retrying Beta pass at 3300.0
Retrying Beta pass at 3400.0
Retrying Beta pass at 3500.0
Retrying Beta pass at 3600.0
Retrying Beta pass at 3700.0
Retrying Beta pass at 3800.0
Retrying Beta pass at 3900.0
Retrying Beta pass at 4000.0
Retrying Beta pass at 4100.0
Retrying Beta pass at 4200.0
Retrying Beta pass at 4300.0
Retrying Beta pass at 4400.0
Retrying Beta pass at 4500.0
Retrying Beta pass at 4600.0
Retrying Beta pass at 4700.0
Retrying Beta pass at 4800.0
Retrying Beta pass at 4900.0
Retrying Beta pass at 5000.0
No path found in beta pass
WARNING [-7324] StepBack: Bad data or over pruning
in /usr/local/HTS-2.1/bin/HERest
My questions are:
Would eliminating this error help increase the quality of synthesized speech?
Where do I turn the pruning off?
Any help would be appreciated,
--
Best regards,
Tomasz Kuczmarski
e-mail: faqster@xxxxxxxxx
skype : faqster
- Follow-Ups
-
- [hts-users:02016] Re: Poor voice quality & embedded reestimation (monophone) errors, Tomasz Kuczmarski