Max Fidelity for wave files with Praat


Hugh Paterson III
 

Greetings,

Does anyone know what the max fidelity that praat can process on wav files?

Wav files can have different audio fidelities for example archival standards for digitizing physical media files are generally 96,000Hz by 24bit. While direct born digital audio recordings are often done at 48,000Hz and 24bit. Other sampling rates and bit rate combinations are possible. I'm looking at the compatibility of archival standards — industry wide— and phonetic software such as praat... I'm also looking at if down sampling is needed in these cases and what might be lost. But I can't really find the documentation for Praat's max fidelity capability.

thanks,

Hugh Paterson III


Boersma Paul
 

Praat will read 24-bit or 32-bit WAV files at sample rates of 96 kHz or 192 kHz just fine. When you view these files, Praat may *play* them at lower sample rates or bit rates, but that doesn't affect the data.

If you edit such objects, Praat can save them again as WAV files in their own sample rate, and you can choose from 16-bit, 24-bit, and 32-bit digitization in the Save menu.

Internally, Praat stores sounds in 64-bit floating-point format, which has 54 bits precision for values between ±0.5 and ±1.0. Praat is mostly also fine with sample rates of e.g. 0.001 Hz or 3.8 GHz.

The maximum fidelity for *recording* with Praat ("Record mono/stereo Sound...") on most computers is 16-bit at 96 kHz.

On 21 Nov 2022, at 22:27, Hugh Paterson III via groups.io <sil.linguist@...> wrote:

Greetings,

Does anyone know what the max fidelity that praat can process on wav files?

Wav files can have different audio fidelities for example archival standards for digitizing physical media files are generally 96,000Hz by 24bit. While direct born digital audio recordings are often done at 48,000Hz and 24bit. Other sampling rates and bit rate combinations are possible. I'm looking at the compatibility of archival standards — industry wide— and phonetic software such as praat... I'm also looking at if down sampling is needed in these cases and what might be lost. But I can't really find the documentation for Praat's max fidelity capability.

thanks,

Hugh Paterson III

_____

Paul Boersma
Professor of Phonetic Sciences
University of Amsterdam
Spuistraat 134, room 632
1012VB Amsterdam, The Netherlands
http://www.fon.hum.uva.nl/paul/