audacia/audacia
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

EXPERIMENTAL_VOICE_DETECTION: compile, use const, more careful casts

Browse Source
This commit is contained in:
Paul Licameli 2017-01-04 17:36:05 -05:00
parent 4c6306e6d6
commit 1e8f0e5a59
4 changed files with 91 additions and 69 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
src/VoiceKey.cpp
View File

@ -85,7 +85,8 @@ VoiceKey::~VoiceKey()
//Move forward to find an ON region.
sampleCount VoiceKey::OnForward (WaveTrack & t, sampleCount start, sampleCount len)
sampleCount VoiceKey::OnForward (
const WaveTrack & t, sampleCount start, sampleCount len)
{
if((mWindowSize) >= (len + 10).as_double() ){
@ -236,7 +237,8 @@ sampleCount VoiceKey::OnForward (WaveTrack & t, sampleCount start, sampleCount l
}
//Move backward from end to find an ON region.
sampleCount VoiceKey::OnBackward (WaveTrack & t, sampleCount end, sampleCount len)
sampleCount VoiceKey::OnBackward (
const WaveTrack & t, sampleCount end, sampleCount len)
{
@ -374,7 +376,8 @@ sampleCount VoiceKey::OnBackward (WaveTrack & t, sampleCount end, sampleCount le
//Move froward from the start to find an OFF region.
sampleCount VoiceKey::OffForward (WaveTrack & t, sampleCount start, sampleCount len)
sampleCount VoiceKey::OffForward (
const WaveTrack & t, sampleCount start, sampleCount len)
{
if((mWindowSize) >= (len + 10).as_double() ){
@ -509,7 +512,8 @@ sampleCount VoiceKey::OffForward (WaveTrack & t, sampleCount start, sampleCount
//Move backward from the end to find an OFF region
sampleCount VoiceKey::OffBackward (WaveTrack & t, sampleCount end, sampleCount len)
sampleCount VoiceKey::OffBackward (
const WaveTrack & t, sampleCount end, sampleCount len)
{
@ -648,7 +652,8 @@ sampleCount VoiceKey::OffBackward (WaveTrack & t, sampleCount end, sampleCount l
}
//This tests whether a specified block region is above or below threshold.
bool VoiceKey::AboveThreshold(WaveTrack & t, sampleCount start, sampleCount len)
bool VoiceKey::AboveThreshold(
const WaveTrack & t, sampleCount start, sampleCount len)
{
double erg=0;
@ -729,7 +734,7 @@ void VoiceKey::AdjustThreshold(double t)
//This 'calibrates' the voicekey to noise
void VoiceKey::CalibrateNoise(WaveTrack & t, sampleCount start, sampleCount len)
void VoiceKey::CalibrateNoise(const WaveTrack & t, sampleCount start, sampleCount len)
{
//To calibrate the noise, we need to scan the sample block just like in the voicekey and
//calculate the mean and standard deviation of the test statistics.
@ -835,7 +840,8 @@ void VoiceKey::SetKeyType(bool erg, bool scLow , bool scHigh,
//This might continue over a number of blocks.
double VoiceKey::TestEnergy (WaveTrack & t, sampleCount start, sampleCount len)
double VoiceKey::TestEnergy (
const WaveTrack & t, sampleCount start, sampleCount len)
{
double sum = 1;
@ -876,7 +882,8 @@ void VoiceKey::TestEnergyUpdate (double & prevErg, int len, const float & drop,
}
double VoiceKey::TestSignChanges(WaveTrack & t, sampleCount start, sampleCount len)
double VoiceKey::TestSignChanges(
const WaveTrack & t, sampleCount start, sampleCount len)
{
@ -932,7 +939,8 @@ void VoiceKey::TestSignChangesUpdate(double & currentsignchanges, int len,
}
double VoiceKey::TestDirectionChanges(WaveTrack & t, sampleCount start, sampleCount len)
double VoiceKey::TestDirectionChanges(
const WaveTrack & t, sampleCount start, sampleCount len)
{

20
src/VoiceKey.h
View File

@ -35,16 +35,16 @@ class VoiceKey {
public:
VoiceKey();
~VoiceKey();
sampleCount OnForward (WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OnBackward (WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OffForward (WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OffBackward (WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OnForward (const WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OnBackward (const WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OffForward (const WaveTrack & t, sampleCount start, sampleCount len);
sampleCount OffBackward (const WaveTrack & t, sampleCount start, sampleCount len);
void CalibrateNoise(WaveTrack & t, sampleCount start, sampleCount len);
void CalibrateNoise(const WaveTrack & t, sampleCount start, sampleCount len);
void AdjustThreshold(double t);
bool AboveThreshold(WaveTrack & t, sampleCount start,sampleCount len);
bool AboveThreshold(const WaveTrack & t, sampleCount start,sampleCount len);
void SetKeyType(bool erg, bool scLow, bool scHigh,
bool dcLow, bool dcHigh);
@ -79,9 +79,11 @@ class VoiceKey {
double mSilentWindowSize; //Time in milliseconds of below-threshold windows required for silence
double mSignalWindowSize; //Time in milliseconds of above-threshold windows required for speech
double TestEnergy (WaveTrack & t, sampleCount start,sampleCount len);
double TestSignChanges (WaveTrack & t, sampleCount start, sampleCount len);
double TestDirectionChanges(WaveTrack & t, sampleCount start, sampleCount len);
double TestEnergy (const WaveTrack & t, sampleCount start,sampleCount len);
double TestSignChanges (
const WaveTrack & t, sampleCount start, sampleCount len);
double TestDirectionChanges(
const WaveTrack & t, sampleCount start, sampleCount len);
void TestEnergyUpdate (double & prevErg, int length, const float & drop, const float & add);
void TestSignChangesUpdate(double & currentsignchanges,int length, const float & a1,

112
src/toolbars/TranscriptionToolBar.cpp
View File

@ -393,7 +393,8 @@ void TranscriptionToolBar::SetButton(bool down, AButton* button)
}
}
void TranscriptionToolBar::GetSamples(WaveTrack *t, sampleCount *s0, sampleCount *slen)
void TranscriptionToolBar::GetSamples(
const WaveTrack *t, sampleCount *s0, sampleCount *slen)
{
// GetSamples attempts to translate the start and end selection markers into sample indices
// These selection numbers are doubles.
@ -528,15 +529,16 @@ void TranscriptionToolBar::OnStartOn(wxCommandEvent & WXUNUSED(event))
Track *t = iter.First(); //Make a track
if(t ) {
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
//if(len == 0)
//len = (WaveTrack*)t->GetSequence()->GetNumSamples()-start;
//len = wt->GetSequence()->GetNumSamples()-start;
auto newstart = mVk->OnForward(*(WaveTrack*)t,start,len);
double newpos = newstart / ((WaveTrack*)t)->GetRate();
auto newstart = mVk->OnForward(*wt, start, len);
double newpos = newstart.as_double() / wt->GetRate();
p->SetSel0(newpos);
p->RedrawProject();
@ -561,15 +563,16 @@ void TranscriptionToolBar::OnStartOff(wxCommandEvent & WXUNUSED(event))
SetButton(false, mButtons[TTB_StartOff]);
Track *t = iter.First(); //Make a track
if(t) {
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
//if(len == 0)
//len = (WaveTrack*)t->GetSequence()->GetNumSamples()-start;
//len = wt->GetSequence()->GetNumSamples()-start;
auto newstart = mVk->OffForward(*(WaveTrack*)t,start,len);
double newpos = newstart / ((WaveTrack*)t)->GetRate();
auto newstart = mVk->OffForward(*wt, start, len);
double newpos = newstart.as_double() / wt->GetRate();
p->SetSel0(newpos);
p->RedrawProject();
@ -594,8 +597,9 @@ void TranscriptionToolBar::OnEndOn(wxCommandEvent & WXUNUSED(event))
Track *t = iter.First(); //Make a track
if(t) {
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
if(len == 0)
@ -603,8 +607,8 @@ void TranscriptionToolBar::OnEndOn(wxCommandEvent & WXUNUSED(event))
len = start;
start = 0;
}
auto newEnd = mVk->OnBackward(*(WaveTrack*)t,start+ len,len);
double newpos = newEnd / ((WaveTrack*)t)->GetRate();
auto newEnd = mVk->OnBackward(*wt, start + len, len);
double newpos = newEnd.as_double() / wt->GetRate();
p->SetSel1(newpos);
p->RedrawProject();
@ -630,16 +634,17 @@ void TranscriptionToolBar::OnEndOff(wxCommandEvent & WXUNUSED(event))
Track *t = iter.First(); //Make a track
if(t) {
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
if(len == 0) {
len = start;
start = 0;
}
auto newEnd = mVk->OffBackward(*(WaveTrack*)t,start+ len,len);
double newpos = newEnd / ((WaveTrack*)t)->GetRate();
auto newEnd = mVk->OffBackward(*wt, start + len, len);
double newpos = newEnd.as_double() / wt->GetRate();
p->SetSel1(newpos);
p->RedrawProject();
@ -665,25 +670,27 @@ void TranscriptionToolBar::OnSelectSound(wxCommandEvent & WXUNUSED(event))
TrackList *tl = p->GetTracks();
TrackListIterator iter(tl);
TrackListOfKindIterator iter(Track::Wave, tl);
Track *t = iter.First(); //Make a track
if(t)
{
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
//if(len == 0)
//len = (WaveTrack*)t->GetSequence()->GetNumSamples()-start;
//len = wt->GetSequence()->GetNumSamples()-start;
double rate = ((WaveTrack*)t)->GetRate();
auto newstart = mVk->OffBackward(*(WaveTrack*)t,start,start);
auto newend = mVk->OffForward(*(WaveTrack*)t,start+len,(int)(tl->GetEndTime()*rate));
double rate = wt->GetRate();
auto newstart = mVk->OffBackward(*wt, start, start);
auto newend =
mVk->OffForward(*wt, start + len, (int)(tl->GetEndTime() * rate));
//reset the selection bounds.
p->SetSel0(newstart / rate);
p->SetSel1(newend / rate);
p->SetSel0(newstart.as_double() / rate);
p->SetSel1(newend.as_double() / rate);
p->RedrawProject();
}
@ -705,24 +712,26 @@ void TranscriptionToolBar::OnSelectSilence(wxCommandEvent & WXUNUSED(event))
TrackList *tl = p->GetTracks();
TrackListIterator iter(tl);
TrackListOfKindIterator iter(Track::Wave, tl);
Track *t = iter.First(); //Make a track
if(t)
{
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
//if(len == 0)
//len = (WaveTrack*)t->GetSequence()->GetNumSamples()-start;
double rate = ((WaveTrack*)t)->GetRate();
auto newstart = mVk->OnBackward(*(WaveTrack*)t,start,start);
auto newend = mVk->OnForward(*(WaveTrack*)t,start+len,(int)(tl->GetEndTime()*rate));
//len = wt->GetSequence()->GetNumSamples()-start;
double rate = wt->GetRate();
auto newstart = mVk->OnBackward(*wt, start, start);
auto newend =
mVk->OnForward(*wt, start + len, (int)(tl->GetEndTime() * rate));
//reset the selection bounds.
p->SetSel0(newstart / rate);
p->SetSel1(newend / rate);
p->SetSel0(newstart.as_double() / rate);
p->SetSel1(newend.as_double() / rate);
p->RedrawProject();
}
@ -745,15 +754,16 @@ void TranscriptionToolBar::OnCalibrate(wxCommandEvent & WXUNUSED(event))
AudacityProject *p = GetActiveProject();
TrackList *tl = p->GetTracks();
TrackListIterator iter(tl);
TrackListOfKindIterator iter(Track::Wave, tl);
Track *t = iter.First(); //Get a track
if(t)
{
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
mVk->CalibrateNoise(*((WaveTrack*)t),start,len);
mVk->CalibrateNoise(*wt, start, len);
mVk->AdjustThreshold(3);
mButtons[TTB_StartOn]->Enable();
@ -798,13 +808,14 @@ void TranscriptionToolBar::OnAutomateSelection(wxCommandEvent & WXUNUSED(event))
mVk->AdjustThreshold(GetSensitivity());
AudacityProject *p = GetActiveProject();
TrackList *tl = p->GetTracks();
TrackListIterator iter(tl);
TrackListOfKindIterator iter(Track::Wave, tl);
Track *t = iter.First(); //Make a track
if(t)
{
sampleCount start,len;
GetSamples((WaveTrack*)t, &start,&len);
auto wt = static_cast<const WaveTrack*>(t);
sampleCount start, len;
GetSamples(wt, &start, &len);
//Adjust length to end if selection is null
if(len == 0)
@ -812,12 +823,12 @@ void TranscriptionToolBar::OnAutomateSelection(wxCommandEvent & WXUNUSED(event))
len = start;
start = 0;
}
int lastlen = 0;
sampleCount lastlen = 0;
double newStartPos, newEndPos;
//This is the minumum word size in samples (.05 is 50 ms)
int minWordSize = (int)(((WaveTrack*)t)->GetRate() * .05);
int minWordSize = (int)(wt->GetRate() * .05);
//Continue until we have processed the entire
//region, or we are making no progress.
@ -826,7 +837,7 @@ void TranscriptionToolBar::OnAutomateSelection(wxCommandEvent & WXUNUSED(event))
lastlen = len;
auto newStart = mVk->OnForward(*(WaveTrack*)t,start,len);
auto newStart = mVk->OnForward(*wt, start, len);
//JKC: If no start found then don't add any labels.
if( newStart==start)
@ -843,7 +854,8 @@ void TranscriptionToolBar::OnAutomateSelection(wxCommandEvent & WXUNUSED(event))
//OK, now we have found a NEW starting point. A 'word' should be at least
//50 ms long, so jump ahead minWordSize
auto newEnd = mVk->OffForward(*(WaveTrack*)t,newStart+minWordSize, len);
auto newEnd =
mVk->OffForward(*wt, newStart + minWordSize, len);
//If newEnd didn't move, we should give up, because
// there isn't another end before the end of the selection.
@ -855,8 +867,8 @@ void TranscriptionToolBar::OnAutomateSelection(wxCommandEvent & WXUNUSED(event))
len -= (newEnd - newStart);
//Calculate the start and end of the words, in seconds
newStartPos = newStart / ((WaveTrack*)t)->GetRate();
newEndPos = newEnd / ((WaveTrack*)t)->GetRate();
newStartPos = newStart.as_double() / wt->GetRate();
newEndPos = newEnd.as_double() / wt->GetRate();
//Increment

2
src/toolbars/TranscriptionToolBar.h
View File

@ -129,7 +129,7 @@ class TranscriptionToolBar final : public ToolBar {
void MakeAlternateImages(
teBmps eFore, teBmps eDisabled,
int id, unsigned altIdx);
void GetSamples(WaveTrack *t, sampleCount *s0, sampleCount *slen);
void GetSamples(const WaveTrack *t, sampleCount *s0, sampleCount *slen);
void SetButton(bool newstate, AButton *button);
void RegenerateTooltips() override;