srconv

Sample rate conversion utility

The srconv utility performs high-quality sample rate conversion on audio files using Kaiser-windowed sinc interpolation. It can perform fixed or time-varying rate conversion with adjustable quality settings.

Syntax

srconv [flags] infile

Required options

Option	Description
-r <rate>	Output sample rate in Hz (required)
-R <rate>	Input sample rate in Hz (required if not in header)

Rate control options

One of these must be specified:

Option	Description
-P <ratio>	Pitch transposition ratio (input_rate / output_rate)
-r <rate>	Output sample rate

Quality and processing options

Option	Description
-Q <factor>	Quality factor from 1 (low) to 8 (high), default: 2
-i <filename>	Time-varying ratio control breakpoint file

Output format options

Option	Description
-o <filename>	Output sound filename
-A	Create AIFF format output
-W	Create WAV format output
-J	Create IRCAM format output
-h	No header (raw output)

Sample format options

Option	Description
-c	8-bit signed char
-8	8-bit unsigned char
-a	A-law encoded
-u	μ-law encoded
-s	16-bit short int (default)
-l	32-bit long int
-f	32-bit float

Additional options

Option	Description
-K	Do not generate PEAK chunks
-R	Continuously rewrite header during processing
-H <n>	Heartbeat display: 1=spinner, 2=dots, 3=numeric
-N	Ring bell when complete
-- <file>	Log output to file

Examples

Basic sample rate conversion

Convert 48kHz to 44.1kHz:

srconv -r 44100 input_48k.wav -o output_44k.wav

Downsample with quality control

srconv -r 22050 -Q 4 input.wav -o output.wav

Upsample to higher rate

srconv -r 96000 -Q 6 audio_44k.wav -o audio_96k.wav

Conversion with explicit input rate

srconv -R 48000 -r 44100 rawfile.raw -o output.wav

Using pitch ratio

srconv -P 1.5 input.wav -o output.wav

This transposes pitch up by a perfect fifth (ratio 3/2).

High-quality conversion

srconv -r 44100 -Q 8 hires.wav -o standard.wav

Time-varying conversion

srconv -r 44100 -i ratio.brk input.wav -o output.wav

With ratio.brk containing:

0.0 1.0
2.0 1.5
4.0 1.0

This gradually speeds up and slows down over time.

Quality factor

The -Q parameter controls interpolation quality:

Q value	Filter length	CPU usage	Quality
1	Shortest	Lowest	Low (audible artifacts)
2	Short	Low	Standard (default)
3-4	Medium	Medium	Good (transparent)
5-6	Long	High	Very good
7-8	Longest	Highest	Excellent (archival)

Choosing quality

• Q=1-2: Real-time processing, low CPU
• Q=3-4: General purpose, good quality/speed balance
• Q=5-6: Professional audio production
• Q=7-8: Mastering, archival conversion

Higher Q values use longer filters with better frequency response and less aliasing.

Sample rate combinations

Common conversions

From	To	Ratio	Use case
48000	44100	0.91875	Video to CD
44100	48000	1.08844	CD to video
44100	22050	0.5	Downsampling by 2
22050	44100	2.0	Upsampling by 2
48000	96000	2.0	Standard to high-res
96000	48000	0.5	High-res to standard

Upsampling vs downsampling

Upsampling (increasing sample rate):

• Adds high-frequency space (initially empty)
• Minimal quality loss
• Larger output files

Downsampling (decreasing sample rate):

• Removes high frequencies
• Requires good anti-aliasing filter (higher Q)
• Smaller output files

Time-varying conversion

The -i breakpoint file format:

time1 ratio1
time2 ratio2
...

• Times in seconds
• Ratios specify input_rate/output_rate at each time
• Linear interpolation between breakpoints

Example uses:

• Variable-speed playback effects
• Pitch glides
• Time stretching with pitch change

Technical details

Interpolation method

srconv uses Kaiser-windowed sinc interpolation:

1. Sinc function provides ideal low-pass filter
2. Kaiser window reduces ripple and ringing
3. Window length determined by quality factor
4. Fractional delay achieved by sinc interpolation

Anti-aliasing

For downsampling:

• Low-pass filter at new Nyquist frequency
• Filter prevents aliasing artifacts
• Higher Q provides sharper cutoff
• Transition band width decreases with Q

Passband and stopband

• Passband: Frequencies preserved (less than 0.45 × output_rate)
• Transition: Roll-off region (depends on Q)
• Stopband: Frequencies rejected (greater than 0.5 × output_rate)

Performance considerations

Processing time

Factors affecting speed:

• Quality factor (Q): Higher Q = slower
• File duration: Linear scaling
• Sample rate change: Upsampling slower than downsampling

Memory usage

Typical: 4096-8192 samples per buffer

Memory = buffer_size × channels × 4 bytes × 2 (input + output)

Common issues

Aliasing artifacts

If downsampled audio has harsh, inharmonic artifacts:

• Increase quality factor (-Q 6 or higher)
• Ensure input sample rate is correct

Slow processing

If conversion is too slow:

• Reduce quality factor (try -Q 3)
• Process in smaller segments
• Use batch processing overnight

Incorrect pitch

If output pitch is wrong:

• Verify input sample rate (-R value)
• Check output rate (-r value)
• Ensure ratio calculation is correct

Related utilities

• sndinfo - Check input/output sample rates
• scale - Adjust amplitude after conversion
• mixer - Mix files with different sample rates