Skip to content

Commit

Permalink
20240529
Browse files Browse the repository at this point in the history
  • Loading branch information
BlitherBoom812 committed May 29, 2024
1 parent 4b96259 commit 45eef7e
Show file tree
Hide file tree
Showing 10 changed files with 175 additions and 10 deletions.
82 changes: 74 additions & 8 deletions source/_posts/SolidPhysics.md
Original file line number Diff line number Diff line change
Expand Up @@ -735,7 +735,6 @@ p 区看电子,n 区看空穴:

![1716259893782](../images/SolidPhysics/1716259893782.png)


导带能级差:$\Delta E_C=\chi_1-\chi_2$
价带能级差:$\Delta E_v=(\chi_2+E_{g^2})-(\chi_1+E_{g^1})=E_{g^2}-E_{g^1}-\Delta E_{g^2}$
导带能级差+价带能级差 =带隙宽度差
Expand All @@ -745,8 +744,6 @@ p 区看电子,n 区看空穴:

### 同质结的注入比



总电流:$j= j_n+ j_p= j_s\left ( e^{qV/ K_BT}- 1\right )$

注入比定义:总电流中,电子电流与空穴电流的比例
Expand All @@ -756,8 +753,6 @@ p 区看电子,n 区看空穴:

注入到$n$区的空穴电流密度为:$j_p= q\frac {D_p}{L_p}p_N^0\left ( e^{qV/ k_BT}- 1\right )$



总电流:$j= j_n+ j_p= j_s\left ( e^{qV/ k_BT}- 1\right )$
注入比定义:总电流中,电子电流与空穴电流的比例
正偏压下的电子注入比:
Expand All @@ -768,8 +763,6 @@ $$

提高注入比的办法,提高N型区的施主杂质浓度



![1716261236429](../images/SolidPhysics/1716261236429.png)

结中电子注入比:
Expand All @@ -796,7 +789,6 @@ $$

$$
\hbar\omega\geq E_{_g}
$$

准动量守恒——竖直跃迁
Expand Down Expand Up @@ -840,3 +832,77 @@ $$
![1716263918247](../images/SolidPhysics/1716263918247.png)

![1716263941957](../images/SolidPhysics/1716263941957.png)

![1716863556246](../images/SolidPhysics/1716863556246.png)

欧姆接触

![1716864038557](../images/SolidPhysics/1716864038557.png)

![1716864050264](../images/SolidPhysics/1716864050264.png)

欧姆接触是指金属与半导体的接触,而其接触面的电阻值远小于半导体本身的电阻,不产生明显的附加阻抗,而且不会使半导体内部的平衡载流子浓度发生显著的改变

区分电子和空穴的本质是迁移率的不同。

## 格波

### 一维单原子链

相邻原子间距为

$$
a+(\mu_{n+1}-\mu_n)=a+\delta
$$

相邻原子的作用力

$$
F=-\frac{\partial\nu}{\partial\delta}\approx-\beta\delta
$$

左右两边一个原子对中间原子的作用力

$$
F_{n,n-1}=-\beta\left(\mu_n-\mu_{n-1}\right)\\
F_{n,n+1}=-\beta\left(\mu_n-\mu_{n+1}\right)\\
F_n=\beta(\mu_{n+1}+\mu_{n-1}-2\mu_n)
$$

相邻原子之间的耦合运动方程组

$$
m\ddot{\mu}_n=\beta\left(\mu_{n+1}+\mu_{n-1}-2\mu_n\right)
$$

通过坐标变换简化方程组,得到简谐振动解

$$
\mu_n=Ae^{i(\omega t-qX_n)}
$$

其中 $X_n=na$ 是第 n 个原子的平衡位置

将解代入运动方程:

$$
m{\left(i\omega\right)}^2Ae^{i(\omega t-qX_n)}=\beta Ae^{i(\omega t-qna)}{\left(e^{-iqa}+e^{iqa}-2\right)}
$$

得到色散关系

$$
\omega^2=\frac{2\beta}m\Big(1-\cos\alpha q\Big)=\frac{4\beta}m\sin^2\Bigg(\frac{aq}2\Bigg)
$$

![1716868157804](../images/SolidPhysics/1716868157804.png)

格波有意义的取值只在第一布里渊区,我们只关心格点位置处的取值。

![1716868762789](../images/SolidPhysics/1716868762789.png)

波恩卡门条件下的环状链模型

![1716869179928](../images/SolidPhysics/1716869179928.png)

![1716869156196](../images/SolidPhysics/1716869156196.png)
103 changes: 101 additions & 2 deletions source/_posts/Speech-SP.md
Original file line number Diff line number Diff line change
Expand Up @@ -579,7 +579,6 @@ $$

#### 最佳正交变换 - KL 变换


## 语音信号的参数编码

### 感觉加权滤波器
Expand Down Expand Up @@ -612,7 +611,7 @@ $\gamma$ 为加权因子,在 0-1 之间。

$\gamma=0$ 时变成逆滤波器,其频谱包络的峰值点就是语音谱的谷值点。

> 分析:语音信号是全极点模型产生的,即 $S(z) = GE(z)/A(z)$,与逆滤波器点频谱成反比。
> 分析:语音信号是全极点模型产生的,即 $S(z) = GE(z)/A(z)$,与逆滤波器点频谱成反比。
![1715172645812](../images/Speech-SP/1715172645812.png)

Expand Down Expand Up @@ -690,3 +689,103 @@ $$
接下来搜索 $n_1$ 使得上式最优化。解出 $n_1$ 后即可解出 $g_1$。

接下来一个个求解,每次都要把前面求解过的脉冲折算到零输入响应中,然后求解当前的结果:

## 语音信号修整与综合技术

短时谱分析

$$
\begin{gathered}w[n]=\begin{cases}\neq0&,&0\leq n<L\\0&,&\text{其它}\end{cases}\\X(p,k)=\sum_{n=-\infty}^{+\infty}x[n]w[p-n]e^{-j\frac{2\pi}{N}nk}\\x[n]w[p-n]=\left[\frac{1}{N}\sum_{i=0}^{N-1}X(p,k)e^{j\frac{2\pi}{N}nk}\right]\cdot R[p-n]\end{gathered}
$$

### 利用修正短时谱进行最小方差信号估计

首先计算短时谱

$$
X(p,\omega)=\sum_{m=-\infty}^{+\infty}x[m]w(p-m)e^{-j\omega m}
$$

对短时谱进行修正

$$
Y(p,\omega)=X(p,\omega)H(p,\omega)
$$

需要得到一个“有效”的短时谱,满足:

1. 时域短时段有限长,非零段不超过窗函数 $w(p-n)$的范围
2. 一致性约束,即不同的𝑝时刻 𝑌(𝑝,𝜔)所对应的时域短时段,若有部分重叠,那么在去除分析窗的加权影响后,在重叠的部分,它们应该是相等的。

实际上很难得到上述信号,可以通过最小方差准则逼近

$$
D[\hat{Y}(n,\omega),Y(n,\omega)]=\sum_{m=-\infty}^{+\infty}\frac{1}{2\pi}\int_{-\pi}^{\pi}\left|\hat{Y}(m,\omega)-Y(m,\omega)\right|^{2}d\omega
$$

其中

$$
\hat{Y}(p,\omega)=\sum_{m=-\infty}^{+\infty}\hat{y}[m]w(p-m)e^{-j\omega m}
$$

利用 Parseval 定理

$$
\begin{aligned}&D\big[\hat{Y}(n,\omega),Y(n,\omega)\big]=\sum_{m=-\infty}^{+\infty}\sum_{n=-\infty}^{+\infty}|\hat{y}_{m}(n)-y_{m}(n)|^{2}\\&=\sum_{m=-\infty}^{+\infty}\sum_{n=-\infty}^{+\infty}|\hat{y}(n)w(m-n)-y_{m}(n)|^{2}\end{aligned}
$$

变分法求得最小值

$$
\hat{y}[n]=\frac{\sum_{m=-\infty}^\infty w(m-n)y_m(n)}{\sum_{m=-\infty}^\infty w^2(m-n)}
$$

其中

$$
y_m(n)=\frac{1}{2\pi}\int_{-\pi}^\pi Y(m,\omega)e^{j\omega n}d\omega
$$

应用中,以周期 $T$ 分析

$$
\hat{y}[n]=\frac{\sum_{m=-\infty}^\infty w(mT-n)y_{mT}(n)}{\sum_{m=-\infty}^\infty w^2(mT-n)}
$$

如果 $w(n)$ 取三角窗的平方根

$$
w^2(n)=\begin{cases}1-\frac{|n-T|}{T}&,\quad0\leq n\leq2T\\\\0&,\quad\text{其它}\end{cases}
$$

若 n 此时为奇数,$T = (N-1)/2$,这时$\sum_{m=-\infty}^{\infty}w^{2}(mT-n)=1$,可简化计算,如果频谱无修正,最优解为

$$
\hat{y}(n)=\frac{\sum_{m=-\infty}^\infty w(mT-n)[w(mT-n)x(n)]}{\sum_{m=-\infty}^\infty w^2(mT-n)}=\frac{x(n)\sum_{m=-\infty}^\infty w^2(mT-n)}{\sum_{m=-\infty}^\infty w^2(mT-n)}=x(n)
$$

若改变语音的速度,$p=k\cdot p^{\prime}$,满足$|Y(p,\omega)|=|X(p^{\prime},\omega)|$ 。

$$
y_{p'}(n)=\frac{1}{2\pi}\int_{-\pi}^{\pi}Y(p',\omega)e^{j\omega n}d\omega=w(p'-n)x(n)
$$

让 $\tau_p=p-p^{\prime}=(1-1/k)\cdot p$,

$$
\begin{aligned}&y_p(n)=y_{p^{\prime}}(n-\tau_p)=w(p^{\prime}-n+\tau_p)x(n-\tau_p)\\&=w(p-n)x(n-\tau_p)\end{aligned}
$$

代入重建公式

$$
\hat{y}[n]=\frac{\sum_{m=-\infty}^\infty w^2(mT-n)x(n-\tau_{mT})}{\sum_{m=-\infty}^\infty w^2(mT-n)}
$$

让$\sum_{m=-\infty}^{\infty}W(mT-n)= \sum_{m=-\infty}^{\infty}w^2(mT-n) =1$,

$$
\hat{y}[n]=\sum_{m=-\infty}^{\infty}W(mT-n)x(n-\tau_{mT})
$$
Binary file added source/images/SolidPhysics/1716863556246.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/SolidPhysics/1716864038557.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/SolidPhysics/1716864050264.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/SolidPhysics/1716868157804.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/SolidPhysics/1716868762789.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/SolidPhysics/1716869156196.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/SolidPhysics/1716869179928.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added source/images/Speech-SP/1716988653084.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.

0 comments on commit 45eef7e

Please sign in to comment.