With a mammoth fortnight of experience using Wavepad under my belt, I thought I'd share my musings on a few things about it that rub me the wrong way, though I should preface any criticisms by saying that Wavepad is by far, the best audio-editing software I've come across to date, making one of the most tedious parts of producing small or large audio-books, now a doddle.
As mentioned in my previous post, on my setup, (little RAM, no SSD space available), with large WAV files, Wavepad is slow. As also mentioned, this is compounded by the fact that it has to use a hard-disk as its cache, but given that this drive is relatively fast (90-100MB/s), it seems slower than it should be. This sluggish performance is exacerbated when you try adding other available editing options to the mix.
For instance, Wavepad allows files to be split at 'silences' that it detects in the audio file. I was almost salivating when I first realised that this feature existed, only to end up despondent when I saw just how slow it rendered the Split process. Worse, this, ’splitting at silences' doesn't appear to be done intelligently. While I appreciate that scanning for silences may be inherently slow, the split-files that Wavepad ends up producing, vary widely in length - by as much as 20 min for the 6 hour audio-book (split into 3 pieces) that I created as a test. Given that there is an option for the user to specify the duration of silences that are sought out, as well as the split-number, what ought to happen is that Wavepad should first calculate the split-points, then search forward & backward from these points, until it finds the nearest appropriate silence. What it actually appears to do is to scan the entire file for silences, which is idiotic, particularly given that 'synthetic voices' like that produced by the AT&T speech engine, which Natural Reader (NR) uses, introduce silences of precise duration into their WAV files, which should make end-of-sentence silences, quick & easy to detect. Wavepad's shoddy 'split-on-silences' implementation renders the whole notion impractical, given the "rough", (Wavepad's description), split-file lengths that it produces. At a minimum, Wavepad should place 'moveable' split-points, thereby allowing the user to shift them manually to nearby silences, prior to splitting. Alas, this 'obvious' functionality is only noteworthy by its absence.
Moving on, its next bug-bear involves multi-threading, or in Wavepad’s case, lack thereof. This is one area where, on a multi-core PC, Wavepad should really shine, specifically, simultaneously encoding the core-based number of split-files into their mp3/ogg etc. equivalents. What it actually does is encode, one part at a time, which is ridiculous.
Last on the chopping block is its Audio-tags implementation. After producing a few audio-books with Wavepad, I noticed that when some of them were opened with the VLC media-player, instead of displaying the file by its (renamed) filename, it uses the WAV-file name that NR generated, basically a string of numbers. These I discovered were being saved by NR as audio-tags, and Wavepad was happily writing these non-descripts to the split-files, which VLC then (bizarrely) used as labels during playback, rather than the actual filenames. Thankfully, Wavepad allows Audio-tags to be cleared, so after saving/reloading the WAV-file, things work as they should. But it's just one more unnecessary step that needs doing.
Nevertheless, Wavepad is an excellent piece of software. Another huge plus for me is that it runs almost flawlessly on Linux Wine as well.
As an aside, I've just come across another Windoze app that has left me rather impressed - Atlantis Word Processor. This tiny 5Meg word-processor is fast & feature-rich while also seeming to run flawlessly on Wine. I had been using the snazzy-looking Chinese WPS as my word-processor of choice on Linux for years, but only realised a few days ago that accented letters such as 'é', are unceremoniously deleted from the text-body when the file is loaded... - WTF?!?!? Bewildered by this butchery, I promptly removed it and installed Libre Office 5 instead - only to discover that this had so many (unrelated) problems, that I'd do better to look elsewhere - and that's where Atlantis W.P. came in! Were I to nit-pick, the only thing missing from it is the ability to import ePub texts - particularly since Atlantis already allows exportation in this format. In truth Atlantis W.P. supports relatively few W.P. standards, DOC, DOCX & RTF, being the main ones, but the latter is the only one I'm interested in, the result of my e-book to audio-book conversion preoccupation.
No comments:
Post a Comment