История изменений

Оппа, девелопер мне уже ответил. Говорит, это, типа, фича, а не баг. ☺

Thank you for report and detailed analysis.

The «errors» that you describe are expected when «dither» is enabled. Dithering is a technique commonly employed to reduce quantization noise and extend the usable dynamic range when reducing the bit depth.

By default, Audacity imports audio in 32 bit float format, which allows Audacity to maintain extremely high quality when processing the audio. All audio processing in Audacity is performed in 32 bit float format.

The artificial test of importing a track and exporting it again in the same format is not a typical use for this software. Audacity is a multi-track audio editor, designed for recording, editing and processing audio. The default settings have been chosen to suit that typical use.

For «scientific» measurement applications, the optimum settings would be to use 32 bit float format throughout, including exporting. This would then obviate the need to reduce the sample format and provide full 32 bit float accuracy. Alternatively, dither can be turned off in preferences (Edit menu > Preferences > Quality), though this will of course produce quantization errors if any signal processing is performed.

There is more information about dither and Audacity's use of dither in the Audacity wiki: http://wiki.audacityteam.org/wiki/Dither

Steve

Действительно, если переключить формат на 16-bit int и отключить dithering, то на входе и выходе сигналы идентичны, в том числе при импорте флака.

Оппа, девелопер мне уже ответил. Говорит, это, типа, фича, а не баг. ☺

Thank you for report and detailed analysis.

The «errors» that you describe are expected when «dither» is enabled. Dithering is a technique commonly employed to reduce quantization noise and extend the usable dynamic range when reducing the bit depth.

By default, Audacity imports audio in 32 bit float format, which allows Audacity to maintain extremely high quality when processing the audio. All audio processing in Audacity is performed in 32 bit float format.

The artificial test of importing a track and exporting it again in the same format is not a typical use for this software. Audacity is a multi-track audio editor, designed for recording, editing and processing audio. The default settings have been chosen to suit that typical use.

For «scientific» measurement applications, the optimum settings would be to use 32 bit float format throughout, including exporting. This would then obviate the need to reduce the sample format and provide full 32 bit float accuracy. Alternatively, dither can be turned off in preferences (Edit menu > Preferences > Quality), though this will of course produce quantization errors if any signal processing is performed.

There is more information about dither and Audacity's use of dither in the Audacity wiki: http://wiki.audacityteam.org/wiki/Dither

Steve

Действительно, если переключить формат на 16-bit int и отключить dithering, то на входе и выходе сигналы идентичны.

Оппа, девелопер мне уже ответил. Говорит, это, типа, фича, а не баг. ☺

Thank you for report and detailed analysis.

The «errors» that you describe are expected when «dither» is enabled. Dithering is a technique commonly employed to reduce quantization noise and extend the usable dynamic range when reducing the bit depth.

By default, Audacity imports audio in 32 bit float format, which allows Audacity to maintain extremely high quality when processing the audio. All audio processing in Audacity is performed in 32 bit float format.

The artificial test of importing a track and exporting it again in the same format is not a typical use for this software. Audacity is a multi-track audio editor, designed for recording, editing and processing audio. The default settings have been chosen to suit that typical use.

For «scientific» measurement applications, the optimum settings would be to use 32 bit float format throughout, including exporting. This would then obviate the need to reduce the sample format and provide full 32 bit float accuracy. Alternatively, dither can be turned off in preferences (Edit menu > Preferences > Quality), though this will of course produce quantization errors if any signal processing is performed.

There is more information about dither and Audacity's use of dither in the Audacity wiki: http://wiki.audacityteam.org/wiki/Dither

Steve