extern/Thorsten-Voice
1
0
Fork 1
mirror of https://github.com/thorstenMueller/Thorsten-Voice.git synced 2025-02-17 18:20:55 +01:00
Code Issues Packages Projects Releases Wiki Activity

First samples

Browse Source
This commit is contained in:
Thorsten Mueller 2020-11-12 22:21:35 +01:00
parent a77b919eaf
commit 84f791f106
17 changed files with 53 additions and 64 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

117
docs/audio_comarison.md
View File

@ -1,67 +1,56 @@
> english version below
# Vocoder Vergleich auf Basis des "thorsten" Tacotron 2 Modells
Hier sind Hörproben mit unterschiedlichen Vocodern. Alle gesprochenen Texte (Sample 1 - 4) basieren auf Aufnahmen im Dataset, jedoch nicht auf dem Spektogramm von "ground truth", sondern auf Basis des trainierten Tacotron 2 Modells. Sample 5 ist der Beginn des Märchens "Der Froschkönig" und wurde nicht für das Dataset aufgezeichnet.
## Sätze
* **Sample #01**: Eure Schoko-Bonbons sind sagenhaft lecker!
* **Sample #02**: Eure Tröte nervt.
* **Sample #03**: Europa und Asien zusammengenommen wird auch als Eurasien bezeichnet.
* **Sample #04**: Euer Plan hat ja toll geklappt.
* *Sample #05: "In den alten Zeiten, wo das Wünschen noch geholfen hat, lebte ein König, dessen Töchter waren alle schön ..." (Anfang vom "Froschkönig")*
# Ground truth
Originalaufnahmen aus dem "thorsten" Dataset.
# Audio vergleich
<dl>
<ul>
<li>Sample #01 - Eure Schoko-Bonbons sind sagenhaft lecker!: <audio controls="" preload="none"><source src="../samples/sample01-gt.wav"></audio></li>
<li>Sample #02 - Eure Tröte nervt.: <audio controls="" preload="none"><source src="../samples/sample02-gt.wav"></audio></li>
<li>Sample #03 - Europa und Asien zusammengenommen wird auch als Eurasien bezeichnet.: <audio controls="" preload="none"><source src="../samples/sample03-gt.wav"></audio></li>
<li>Sample #04 - Euer Plan hat ja toll geklappt.: <audio controls="" preload="none"><source src="../samples/sample04-gt.wav"></audio></li>
</ul>
</dl>
<table width="80%">
<thead>
<tr>
<th >Vocoder</th>
<th >RealTimeFactor</th>
<th >Sample #1</th>
<th >Sample #2</th>
<th >Sample #3</th>
<th >Sample #4</th>
<th >Details</th>
</tr>
</thead>
<tbody>
<tr>
<td >Ground truth</td>
<td > --- </td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
</tr>
<tr>
<td >Griffin lim</td>
<td >0.5</td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ></td>
<td ></td>
<td ></td>
<td ></td>
</tr>
<tr>
<td >ParallelWaveGAN</td>
<td >0.6</td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ></td>
<td ></td>
<td ></td>
<td ></td>
</tr>
<tr>
<td >WaveGrad</td>
<td >0.7</td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ></td>
<td ></td>
<td ></td>
<td ></td>
</tr>
<tr>
<td >HifiGAN</td>
<td >1.5</td>
<td ><audio controls="" preload="none"><source src="./sample.wav"></audio></td>
<td ></td>
<td ></td>
<td ></td>
<td ></td>
</tr>
</tbody>
</table>
</dl>
# Griffin Lim
> Details zum Model: (todo: link)
> Tacotron2 + DDC: 460k Schritte trainiert
# ParallelWaveGAN
> Tacotron2 + DDC: 360k Schritte trainiert
> PWGAN Vocoder: 925k Schritte trainiert
<dl>
<ul>
<li>Sample #01 - Eure Schoko-Bonbons sind sagenhaft lecker!:
<audio controls="" preload="none"><source src="../samples/sample01-pwgan.wav"></audio></li>
<li>Sample #02 - Eure Tröte nervt.:
<audio controls="" preload="none"><source src="../samples/sample02-pwgan.wav"></audio></li>
<li>Sample #03 - Europa und Asien zusammengenommen wird auch als Eurasien bezeichnet.:
<audio controls="" preload="none"><source src="../samples/sample03-pwgan.wav"></audio></li>
<li>Sample #04 - Euer Plan hat ja toll geklappt.:
<audio controls="" preload="none"><source src="../samples/sample04-pwgan.wav"></audio></li>
<li>Sample #05 - Anfang vom Froschkönig:
<audio controls="" preload="none"><source src="../samples/sample04-pwgan.wav"></audio></li>
</ul>
</dl>
# WaveGrad
> todo
# HifiGAN
> todo

BIN
samples/original_recording/recorded_sample_01.wav
View File

Binary file not shown.

BIN
samples/original_recording/recorded_sample_02.wav
View File

Binary file not shown.

BIN
samples/original_recording/recorded_sample_03.wav
View File

Binary file not shown.

BIN
samples/original_recording/recorded_sample_04.wav
View File

Binary file not shown.

BIN
samples/original_recording/recorded_sample_05.wav
View File

Binary file not shown.

BIN
samples/original_recording/recorded_sample_06.wav
View File

Binary file not shown.

BIN
samples/original_recording/recorded_sample_07.wav
View File

Binary file not shown.

BIN
samples/sample01-gt.wav Normal file
View File

Binary file not shown.

BIN
samples/sample01-pwgan.wav Normal file
View File

Binary file not shown.

BIN
samples/sample02-gt.wav Normal file
View File

Binary file not shown.

BIN
samples/sample02-pwgan.wav Normal file
View File

Binary file not shown.

BIN
samples/sample03-gt.wav Normal file
View File

Binary file not shown.

BIN
samples/sample03-pwgan.wav Normal file
View File

Binary file not shown.

BIN
samples/sample04-gt.wav Normal file
View File

Binary file not shown.

BIN
samples/sample04-pwgan.wav Normal file
View File

Binary file not shown.

BIN
samples/sample05-pwgan.wav Normal file
View File

Binary file not shown.