Mercurial > hg > cepstral-pitchtracker
diff NoteHypothesis.cpp @ 60:c06fe5350b34
But don't use a mid-stream low-confidence observation to reject a good ongoing hypothesis
author | Chris Cannam |
---|---|
date | Tue, 20 Nov 2012 21:57:49 +0000 |
parents | 82552664d471 |
children | 7ad142c710c6 |
line wrap: on
line diff
--- a/NoteHypothesis.cpp Tue Nov 20 21:31:17 2012 +0000 +++ b/NoteHypothesis.cpp Tue Nov 20 21:57:49 2012 +0000 @@ -98,7 +98,9 @@ // avoid piling up a lengthy sequence of estimates that are // all acceptable but are in total not enough to cause us to // be satisfied - m_state = Rejected; + if (m_pending.empty()) { + m_state = Rejected; + } return false; }