Libvorbis Sample Rate Conversion and Resampling

This article explains how the libvorbis codec library handles audio sample rates during the encoding and decoding processes. You will learn whether libvorbis includes native resampling capabilities or if it relies on external libraries to convert sample rates.

No Native Sample Rate Conversion in Libvorbis

The libvorbis library does not perform sample rate conversion. It is designed strictly as an audio compression and decompression engine. Its core responsibility is to convert raw PCM audio into compressed Vorbis packets (encoding) and to turn those packets back into raw PCM audio (decoding).

Because digital signal processing (DSP) operations like resampling are highly complex and computationally distinct from compression, the developers of Vorbis chose to keep these systems separate. Consequently, libvorbis expects the input sample rate to match the desired output sample rate of the encoded stream.

Handling Resampling During Encoding

When encoding audio with libvorbis, you must configure the encoder to match the sample rate of your input PCM data. For example, if you feed 44.1 kHz PCM data into the encoder, you must initialize the vorbis_info struct to 44.1 kHz. The resulting Vorbis stream will be flagged and encoded at 44.1 kHz.

If you need to change the sample rate during the encoding process (for example, downsampling a 96 kHz studio master to a 44.1 kHz Ogg Vorbis file), you must perform this conversion before the audio data reaches libvorbis. Developers typically achieve this by piping the audio through an external resampling library. Popular choices include:

• SoX Resampler library (libsoxr): A high-quality, fast resampling library.
• Secret Rabbit Code (libsamplerate): A widely used library offering various converter types ranging from high-quality sinc interpolation to fast linear interpolation.
• SpeexDSP: A lightweight resampler often used in embedded or real-time systems.

Handling Resampling During Decoding

During decoding, libvorbis reads the Ogg Vorbis stream header, identifies the original sample rate, and outputs PCM audio at that exact rate. If a Vorbis file was encoded at 32 kHz, libvorbis will output 32 kHz PCM audio.

If the playback hardware or the audio subsystem (such as Windows WASAPI, macOS CoreAudio, or Linux ALSA) requires a specific sample rate (like 48 kHz), the decoding application must handle the conversion. The application must take the decoded PCM buffer from libvorbis and pass it through an external resampling library or the operating system’s native audio mixer before sending it to the audio hardware.