发烧论坛

注册

 

发新话题 回复该主题

(PC HiFi)HQPlayer的回放设置。 [复制链接]

1#
当下HQPlayer播放软件非常火,虽然这个软件还不太成熟稳定,但他给我们带来听觉感受是前所未有的。由于这个软件涉及许多音频专业术语,所以一直没有汉化版本。给广大烧友带来诸多不便。在我的关于PC HiFi交流的贴中结识HappyWater 兄。深感其专业英文功底深厚,抱着试试看的心态,给他发了短信请其翻译HQPlayer英文说明书中关于回放设置的说明。不想HappyWater兄深夜迅速翻译回复,使我备受感动。也在HappyWater兄回复中感到其扎实丰富的专业知识和严谨的作风。征得其同意特开此贴与广大烧友交流。也希望广大烧友多提宝贵建议。
HQPlayer关于回放设置的E文如下:
none ~~~No sample rate conversion happens.Only sample depth is
changed as needed.

IIR ~~~~This is analog-sounding filter,especially suitable for recordings
containing strong transients,long post-echo is a side effect(not
usually audible due to masking).A really steep IIR filter is used.
This filter type is similar to analog filters and has no pre-echo,but
has a long post-echo.Small amount of pass-band ripple is also
present.IIR filter is applied in time-domain.This filter performs
better when resampling with only factor of 2 or 3 and somewhat
worse at factor of 4 and higher.

FIR ~~~~Typical“oversampling”digital filter,generally suitable for most
uses(slight pre-and post-echo),but best on classical music
recorded in a real world acoustic environment such as concert hall.
This is the most ordinary filter type,usually present in hardware.
This filter is applied in time-domain.It has average amount of pre-
and post-echo.

asymFIR ~~~~Asymmetric FIR,good for jazz/blues,and other music containing
transients recorded in real world acoustic environment.Otherwise
same as FIR,but with a shorter pre-echo and longer post-echo.
Modifies phase response,but not as much as minimum phase
FIR.

minphaseFIR ~~~~Minimum phase FIR,good for pop/rock/electronic music containing
strong transients such as drums and percussion and where
recording is made in a studio using multi-track equipment.No pre-
echo,but somewhat long post-echo.

FFT ~~~~Technically good steep“brickwall”filter,but might have some side
effects(pre-echo)on material containing strong transients.This
filter is similar to FIR,but it is applied in frequency-domain and is
quite efficient from performance point of view while having rather
long impulse response.

poly-sinc/
better space~~~
Linear phase polyphase sinc filter.Very high quality linear phase
resampling filter,can perform most of the typical conversion ratios.
Good phase response,but has some amount of pre-echo.See
“FIR”for further details.

poly-sinc-mp/
better transients~~~~
Minimum phase polyphase sinc filter,otherwise similar to poly-
sinc.Altered phase response,but no pre-echo.See
“minphaseFIR”for further details.

poly-sinc-shrt~~~~ Otherwise similar as poly-sinc,but shorter pre-and post-echos at
the expense of filtering quality(not as sharp roll-off and reduced
stop-band attenuation).

poly-sinc-shrt-mp~~~~ Minimum phase variant of poly-sinc-shrt.Otherwise similar to poly-
sinc-mp,but shorter post-echo.Most optimal transient
reproduction.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
sinc~~~ This is a special type of filter,slightly similar to FIR,but with a
possibility of asynchronous operation for conversions from any
rate to any other rate.Computationally heavy.

polynomial-1~~~~Polynomial interpolation.Most natural polynomial interpolation for
audio.Only two samples of pre-and post-echo.Frequency
response rolls off slowly in the top octave.Poor stop-band
rejection and will thus leak fairly high amount of ultrasonic noise.
These type of filters are sometimes referred to as“non-ringing”by
some manufacturers.Not recommended.

polynomial-2 ~~~~Similar to polynomial-1,but higher stop-band rejection with the
expense of being a bit less natural for audio.Not recommended.

minringFIR ~~~~~Minimum ringing FIR.Uses special algorithm to create a linear-
phase filter that minimizes amount of ringing while providing better
frequency-response and attenuation than polynomial interpolators.
Performance and ringing is between polynomial and poly-sinc-
short.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
none~~~~ No noise-shaping or dithering,only rounding.Mostly suitable for 32-bit
output hardware where quantization error is buried in thermal noise.Not
recommended.

NS1~~~ Simple first order noise-shaping.Sample values are rounded and the
quantization error is shaped such way that the error energy is pushed to
the higher frequencies.Suitable mostly for 176.4/192 kHz upsampling.

NS4~~~ Fourth order noise-shaping.Similar in shape as“shaped”dither.
Suitable for all rates equal or higher than 88.2 kHz.

NS5 ~~Fifth order noise-shaping.Fairly aggressive noise-shaping designed for
8x and 16x rates(352.8/384/705.6/768 kHz).Not recommended for
rates below 192 kHz.(Especially good for PCM1704 at those highest
rates.)

NS9~~~~ Ninth order noise-shaping.Very aggressive noise-shaping designed
especially for 4x rates(176.4/192 kHz)and recommended for these
rates.(Especially good for older 16-bit 4x rate capable multibit-DACs
like TDA154x etc.)

RPDF~~ Rectangular Probability Density Function.White noise dither.
Computationally light weight,but only suitable for 24-bit or higher
output hardware.

TPDF~~~ Triangular Probability Density Function.This is the industry standard
simple dither mechanism.Suitable for any rate and recommended if
playback rate is 44.1/48 kHz.Recommended for general purpose use.

Gauss1 ~~~Gaussian Probability Density Function.Very high quality flat frequency
dither recommended for rates at or below 96 kHz where noise-shaping
is not suitable.

shaped~~~~ Shaped dither.Noise used in this dither has shaped frequency
distribution to lower audibility of the dither noise.Suitable for playback
rates of 88.2/96 kHz,or higher.

下面是HappyWater兄回复中的表述和翻译全文:
“其实我觉得或许可以开贴讨论一下这份翻译,让大家讨论下该怎么润色。
目的当然是润色出一份大家都能看得懂,都能接受的翻译版本。
或许还可以让版主长期置顶这个讨论帖,直到讨论出一个大家都能接受的版本。

可以看到,原文中出现了不少专有名词缩写,为了做到“信”和“达”,我只能按照这些词在国内约定俗成的翻译,但这不一定是广大烧友所习惯的。
因为不了解广大烧友们的习惯,限于鄙人水平,我只能尽力做到“信”和“达”,至于“雅”还是要靠大家的智慧啊。 ”

“none 不做重采样(改变采样率)。只有采样深度会根据需要而改变。

IIR(无限冲击响应滤波器) 这是一个模仿模拟滤波器声效的滤波器,尤其适合于当录音包含大振幅的瞬态(transient)时,存在的副作用是有很长的回声(由于遮蔽效应,通常是觉察不到的)。这里采用的是一个非常陡峭的IIR滤波器。这种滤波器和模拟滤波器很相似,都没有前兆(即,都是因果系统),但是有很长的回音。同时,存在少量通带波纹。IIR滤波器在时域中实现。当进行重采样时,这种滤波器在二阶和三阶的性能会比四阶和更高阶的表现更好。

FIR(有限冲击响应滤波器) 典型的过采样滤波器,基本上可以用于任何滤波需要(少量前回音和后回音),但是,用在从实际声学环境中(例如音乐演出大厅)录制的古典音乐上效果最好。这是最普通的滤波器种类,通常以硬件实现。该种滤波器于时域中实现。它的前回音和后回音水平属于平均水平。

asymFIR(不对称FIR) 不对称FIR,适用于爵士乐和蓝调音乐,以及其他从实际声学环境中录制的并且包含较大振幅瞬态(transient)的音乐。其他方面与FIR一样,不过前回音会小些而后回音会大些。修饰相位响应,但是不如最小相位FIR(minphaseFIR)那样重。

minphaseFIR(最小相位FIR) 最小相位FIR,适合于流行乐/摇滚/电子乐等包含大振幅瞬态(transient)(例如鼓和打击类乐器)的并且是使用工作室用多轨道混音设备制作的音乐录音。没有前回音,但是有一定长度的后回音。

FFT(快速傅里叶变换) 技术上优越的(通带阻带过渡)锋利的“砖墙”滤波器,但是当录音材料包含有大振幅的瞬态时,可能会有一些副作用(前回音)。这种滤波器和FIR相似,但是是在频域实现的,而且,从性能角度来说非常高效,尤其是在冲击响应很长的时候。

poly-sinc/better space(多相位Sinc函数/优化空间感,译注:sinc函数在大陆的习惯称呼为Sa函数,读音“萨函数”,以下都称“萨函数”具体表达式百度就有,不赘述)线性相位多相位萨函数滤波器。非常高质量的线性相位重采样滤波器,能用于大多数典型转换比例(原采样率和目标采样率的比值)。良好的相位响应,不过,有一定的前回音。参见FIR一节。

poly-sinc-mp/better transients(最小相位多相位萨函数/优化瞬态)最小相位多相位萨函数滤波器,除了相位变化最小以外,其他方面都和多相位萨函数滤波器相似。转换相位响应,但是没有前回音。参见最小相位FIR。

poly-sinc-shrt(短回音-多相位萨函数) 其他方面都和多相位萨函数滤波器一样,不同之处在于前回音和后回音都要短,代价是牺牲了一定的滤波质量(并不是“硬滚降”而且降低了阻带衰减)。

poly-sinc-shrt-mp(最小相位短回音多相位萨函数滤波器) 相位失真最小的“短回音-多相位萨函数”。其他方面都和“短回音-多相位萨函数”没啥区别,不过后回音更短。瞬态的再生效果(比之于其它各个滤波器来说)是最为理想的。

sinc(萨函数滤波) 这是一种典型的滤波器,和FIR有点相似,但是可能异步地实现任何频率之间的转换操作。计算量很大。

Polynomial-1 (多项式滤波器 1) 多项式插值法。最自然的音频多项式插值法。前回音和后回音都只有两个采样的长度。在最高频率附近,频响缓慢滚降。阻带衰减性能很弱,因而产生一定量的超声波噪声泄露(译注:也就是混频现象)。这些滤波器有些时候被一些厂家用于“消哨响”(哨响:拿话筒对着喇叭时发出的那种尖锐声音)。不推荐使用。

Polynomial-2 (多项式滤波器 2) 与多项式1相似,但是有更强的阻带衰减能力,但同时又会让声音听起来有些不太自然。不推荐使用。

minringFIR(最小哨响FIR) 最小哨响FIR。利用特殊算法实现一个线性相位滤波器,该滤波器在达到比多项式插值法更好的频响曲线和阻带衰减的同时会把哨响压制到最低水平。计算性能(运算负荷)和哨响介于“多项式插值滤波”和“短回音多相位萨函数滤波”之间。

none 不做噪声整形或抖动,仅仅是四舍五入。最适合的是用于输出32位音频的硬件,因为这种情况下,量化偏差已经被热噪声所掩埋。不推荐使用。

NS1(噪声整形1) 简单的噪声整形算法。采样值被四舍五入,而后修正量化错误,把误差信号的能量推到高频带去。最适用的情况是176.4/192KHz升频。

NS4(噪声整形4) 第四个噪声整形。形状上和整形后的抖动相似。适用于所有大于等于88.2KHz的采样率。

NS5(噪声整形5) 第五个噪声整形。面向未来的噪声整形算法,设计用在8倍和16倍速率(352.8/384/705.6/768KHz)。不推荐使用在192KHz以下的采样率上(尤其适用于工作在其最高速率的 PCM1704 数模转换器上)。

NS9(噪声整形9) 第九个噪声整形算法。面向高端应用的噪声整形算法,尤其用于4倍速率(176.4/192KHz)并且推荐使用在这些频率上。(尤其适用于老式16位4倍率兼容多比特DAC,例如TDA154x之类的。)

RPDF(矩形概率密度函数) 矩形概率密度函数。白噪声抖动。对计算机的负担较低,但只适用于24bit或者更高量化深度的输出硬件上。

TPDF(三角概率密度函数) 三角概率密度函数。这是工业标准的简单抖动机制。适用于任何采样速率并且推荐用于44.1/48KHz的回放采样率。推荐为通用算法。

Gauss1(高斯1,译注:就是瞬间算出了从一加到一百的那个神童 高斯。)高斯概率密度函数(译注:也就是所谓的“正态分布”,参见百度百科“高尔顿板”)。非常高质量的平坦频率抖动。推荐用于小于等于96KHz的采样速率情况,这些速率下,前面几种噪声整形都不太合适。

shaped(整形过) 整形过的抖动。在这种抖动算法中采用的噪声的频谱被整形到了更低的可闻度。适用于回放采样率在88.2/96KHz或者更高的场合。”
最后编辑truemusic 最后编辑于 2012-06-13 18:19:56
分享 转发
玩的是乐趣 享受的是音乐。
TOP
发新话题 回复该主题