Hello,
I want to detect a frequency of my signal. My signal is "well behaving", therefore variable "freq" at stat.c line 285 seems to contain a correct and highly precise value of the frequency of my signal. However, on the next line (stat.c line 286) type conversion (int) rounds that frequency down resulting in one Hertz too little value in half of the cases. e.g.
> sox -n /tmp/tmp.wav synth 1 sine 100.01 stat // gives 100Hz. OK.
> sox -n /tmp/tmp.wav synth 1 sine 99.99 stat // gives 99Hz. should be 100Hz, right
Line 285 in stat.c
is freq = sqrt(stat->dsum2/stat->sum2)*effp->in_signal.rate/(M_PI*2);
should be: freq = 0.5 + sqrt(stat->dsum2/stat->sum2)*effp->in_signal.rate/(M_PI*2);
Or, more preferrably line 286 in stat.c
is fprintf(stderr, "Rough frequency: %12d\n", (int)freq);
should be fprintf(stderr, "Rough frequency: %14.2f\n", freq);
I want to detect a frequency of my signal. My signal is "well behaving", therefore variable "freq" at stat.c line 285 seems to contain a correct and highly precise value of the frequency of my signal. However, on the next line (stat.c line 286) type conversion (int) rounds that frequency down resulting in one Hertz too little value in half of the cases. e.g.
> sox -n /tmp/tmp.wav synth 1 sine 100.01 stat // gives 100Hz. OK.
> sox -n /tmp/tmp.wav synth 1 sine 99.99 stat // gives 99Hz. should be 100Hz, right
Line 285 in stat.c
is freq = sqrt(stat->dsum2/stat->sum2)*effp->in_signal.rate/(M_PI*2);
should be: freq = 0.5 + sqrt(stat->dsum2/stat->sum2)*effp->in_signal.rate/(M_PI*2);
Or, more preferrably line 286 in stat.c
is fprintf(stderr, "Rough frequency: %12d\n", (int)freq);
should be fprintf(stderr, "Rough frequency: %14.2f\n", freq);
Or is the some other effect in sox that could be used for "high precision" frequency detection of a signal?
regards, Mikko O._______________________________________________ Sox-users mailing list Sox-users@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/sox-users
